diff --git a/CMakeLists.txt b/CMakeLists.txt
index 7fd78cac1..583fb8afe 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -226,10 +226,7 @@ endif()
 
 
 # We don't need exceptions, disable them to save a bit of code size.
-# Except xbyak still uses them :(
-if(NOT ("${CPU_ARCH}" STREQUAL "x86" OR "${CPU_ARCH}" STREQUAL "x64"))
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fno-exceptions")
-endif()
+set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fno-exceptions")
 
 
 # Write binaries to a seperate directory.
diff --git a/dep/xbyak/COPYRIGHT b/dep/xbyak/COPYRIGHT
index 66b6ea55d..9c0acf844 100644
--- a/dep/xbyak/COPYRIGHT
+++ b/dep/xbyak/COPYRIGHT
@@ -25,23 +25,3 @@ INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
 CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
 ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
 THE POSSIBILITY OF SUCH DAMAGE.
------------------------------------------------------------------------------
-ソースコード形式かバイナリ形式か、変更するかしないかを問わず、以下の条件を満た
-す場合に限り、再頒布および使用が許可されます。
-
-ソースコードを再頒布する場合、上記の著作権表示、本条件一覧、および下記免責条項
-を含めること。
-バイナリ形式で再頒布する場合、頒布物に付属のドキュメント等の資料に、上記の著作
-権表示、本条件一覧、および下記免責条項を含めること。
-書面による特別の許可なしに、本ソフトウェアから派生した製品の宣伝または販売促進
-に、著作権者の名前またはコントリビューターの名前を使用してはならない。
-本ソフトウェアは、著作権者およびコントリビューターによって「現状のまま」提供さ
-れており、明示黙示を問わず、商業的な使用可能性、および特定の目的に対する適合性
-に関する暗黙の保証も含め、またそれに限定されない、いかなる保証もありません。
-著作権者もコントリビューターも、事由のいかんを問わず、 損害発生の原因いかんを
-問わず、かつ責任の根拠が契約であるか厳格責任であるか（過失その他の）不法行為で
-あるかを問わず、仮にそのような損害が発生する可能性を知らされていたとしても、
-本ソフトウェアの使用によって発生した（代替品または代用サービスの調達、使用の
-喪失、データの喪失、利益の喪失、業務の中断も含め、またそれに限定されない）直接
-損害、間接損害、偶発的な損害、特別損害、懲罰的損害、または結果損害について、
-一切責任を負わないものとします。
diff --git a/dep/xbyak/readme.md b/dep/xbyak/readme.md
index 990188f33..f0ebdb56e 100644
--- a/dep/xbyak/readme.md
+++ b/dep/xbyak/readme.md
@@ -1,443 +1,80 @@
 
-Xbyak 5.41 ; JIT assembler for x86(IA32), x64(AMD64, x86-64) by C++
-=============
+# Xbyak 6.73 [![Badge Build]][Build Status]
 
-Abstract
--------------
+*A C++ JIT assembler for x86 (IA32), x64 (AMD64, x86-64)*
 
-This is a header file which enables dynamically to assemble x86(IA32), x64(AMD64, x86-64) mnemonic.
+## Menu
 
-Feature
--------------
-header file only
-you can use Xbyak's functions at once if xbyak.h is included.
+- [Install]
+- [Usage]
+- [Changelog]
 
-### Supported Instructions Sets
+## Abstract
 
-MMX/MMX2/SSE/SSE2/SSE3/SSSE3/SSE4/FPU(*partial*)/AVX/AVX2/FMA/VEX-encoded GPR/AVX-512
+Xbyak is a C++ header library that enables dynamically to assemble x86(IA32), x64(AMD64, x86-64) mnemonic.
 
-### Supported OS
-
-* Windows Xp, Vista, Windows 7(32bit, 64bit)
-* Linux(32bit, 64bit)
-* Intel Mac OSX
-
-### Supported Compilers
-
-* Visual Studio C++ VC2012 or later
-* gcc 4.7 or later
-* clang 3.3
-* cygwin gcc 4.5.3
-* icc 7.2
-
->Note: Xbyak uses and(), or(), xor(), not() functions, so "-fno-operator-names" option is required on gcc.
-Or define XBYAK_NO_OP_NAMES and use and_(), or_(), xor_(), not_() instead of them.
-and_(), or_(), xor_(), not_() are always available.
-
-Install
--------------
-
-The following files are necessary. Please add the path to your compile directories.
-
-* xbyak.h
-* xbyak_mnemonic.h
-
-Linux:
-
-    make install
-
-These files are copied into /usr/local/include/xbyak
-
-New Feature
--------------
-
-Add support for AVX-512 instruction set.
-
-Syntax
--------------
-
-Make Xbyak::CodeGenerator and make the class method and get the function
-pointer by calling cgetCode() and casting the return value.
-
-    NASM              Xbyak
-    mov eax, ebx  --> mov(eax, ebx);
-    inc ecx           inc(ecx);
-    ret           --> ret();
-
-### Addressing
-
-    (ptr|dword|word|byte) [base + index * (1|2|4|8) + displacement]
-                          [rip + 32bit disp] ; x64 only
-
-    NASM                   Xbyak
-    mov eax, [ebx+ecx] --> mov (eax, ptr[ebx+ecx]);
-    test byte [esp], 4 --> test (byte [esp], 4);
-
-
-How to use Selector(Segment Register)
-
->Note: Segment class is not derived from Operand.
-
-```
-mov eax, [fs:eax] --> putSeg(fs); mov(eax, ptr [eax]);
-mov ax, cs        --> mov(ax, cs);
-```
-
->you can use ptr for almost memory access unless you specify the size of memory.
-
->dword, word and byte are member variables, then don't use dword as unsigned int, for example.
-
-### AVX
-
-    vaddps(xmm1, xmm2, xmm3); // xmm1 <- xmm2 + xmm3
-    vaddps(xmm2, xmm3, ptr [rax]); // use ptr to access memory
-    vgatherdpd(xmm1, ptr [ebp+123+xmm2*4], xmm3);
-
-*Remark*
-The omitted destination syntax as the following ss disabled.
-```
-    vaddps(xmm2, xmm3); // xmm2 <- xmm2 + xmm3
-```
-define `XBYAK_ENABLE_OMITTED_OPERAND` if you use it for backward compatibility.
-But the newer version will not support it.
-
-### AVX-512
+The pronunciation of Xbyak is `kəi-bja-k`.
+It is named from a Japanese word [開闢](https://translate.google.com/?hl=ja&sl=ja&tl=en&text=%E9%96%8B%E9%97%A2&op=translate), which means the beginning of the world.
 
-```
-vaddpd zmm2, zmm5, zmm30                --> vaddpd(zmm2, zmm5, zmm30);
-vaddpd xmm30, xmm20, [rax]              --> vaddpd(xmm30, xmm20, ptr [rax]);
-vaddps xmm30, xmm20, [rax]              --> vaddps(xmm30, xmm20, ptr [rax]);
-vaddpd zmm2{k5}, zmm4, zmm2             --> vaddpd(zmm2 | k5, zmm4, zmm2);
-vaddpd zmm2{k5}{z}, zmm4, zmm2          --> vaddpd(zmm2 | k5 | T_z, zmm4, zmm2);
-vaddpd zmm2{k5}{z}, zmm4, zmm2,{rd-sae} --> vaddpd(zmm2 | k5 | T_z, zmm4, zmm2 | T_rd_sae);
-                                            vaddpd(zmm2 | k5 | T_z | T_rd_sae, zmm4, zmm2); // the position of `|` is arbitrary.
-vcmppd k4{k3}, zmm1, zmm2, {sae}, 5     --> vcmppd(k4 | k3, zmm1, zmm2 | T_sae, 5);
+## Feature
 
-vaddpd xmm1, xmm2, [rax+256]            --> vaddpd(xmm1, xmm2, ptr [rax+256]);
-vaddpd xmm1, xmm2, [rax+256]{1to2}      --> vaddpd(xmm1, xmm2, ptr_b [rax+256]);
-vaddpd ymm1, ymm2, [rax+256]{1to4}      --> vaddpd(ymm1, ymm2, ptr_b [rax+256]);
-vaddpd zmm1, zmm2, [rax+256]{1to8}      --> vaddpd(zmm1, zmm2, ptr_b [rax+256]);
-vaddps zmm1, zmm2, [rax+rcx*8+8]{1to16} --> vaddps(zmm1, zmm2, ptr_b [rax+rcx*8+8]);
-vmovsd [rax]{k1}, xmm4                  --> vmovsd(ptr [rax] | k1, xmm4);
+- header file only
+- Intel/MASM like syntax
+- fully support AVX-512
 
-vcvtpd2dq xmm16, oword [eax+33]         --> vcvtpd2dq(xmm16, xword [eax+33]); // use xword for m128 instead of oword
-                                            vcvtpd2dq(xmm16, ptr [eax+33]); // default xword
-vcvtpd2dq xmm21, [eax+32]{1to2}         --> vcvtpd2dq(xmm21, ptr_b [eax+32]);
-vcvtpd2dq xmm0, yword [eax+33]          --> vcvtpd2dq(xmm0, yword [eax+33]); // use yword for m256
-vcvtpd2dq xmm19, [eax+32]{1to4}         --> vcvtpd2dq(xmm19, yword_b [eax+32]); // use yword_b to broadcast
+**Note**:
+Use `and_()`, `or_()`, ... instead of `and()`, `or()`.
+If you want to use them, then specify `-fno-operator-names` option to gcc/clang.
 
-vfpclassps k5{k3}, zword [rax+64], 5    --> vfpclassps(k5|k3, zword [rax+64], 5); // specify m512
-vfpclasspd k5{k3}, [rax+64]{1to2}, 5    --> vfpclasspd(k5|k3, xword_b [rax+64], 5); // broadcast 64-bit to 128-bit
-vfpclassps k5{k3}, [rax+64]{1to4}, 5    --> vfpclassps(k5|k3, xword_b [rax+64], 5); // broadcast 32-bit to 128-bit
-```
-Remark
-* k1, ..., k7 are new opmask registers.
-* use `| T_z`, `| T_sae`, `| T_rn_sae`, `| T_rd_sae`, `| T_ru_sae`, `| T_rz_sae` instead of `,{z}`, `,{sae}`, `,{rn-sae}`, `,{rd-sae}`, `,{ru-sae}`, `,{rz-sae}` respectively.
-* `k4 | k3` is different from `k3 | k4`.
-* use `ptr_b` for broadcast `{1toX}`. X is automatically determined.
-* specify xword/yword/zword(_b) for m128/m256/m512 if necessary.
+### Derived Projects
+- [Xbyak_aarch64](https://github.com/fujitsu/xbyak_aarch64/) : for AArch64
+- [Xbyak_riscv](https://github.com/herumi/xbyak_riscv) : for RISC-V
 
-### Label
+### News
 
-    L("L1");
-      jmp ("L1");
+- add amx_fp16/avx_vnni_int8/avx_ne_convert/avx-ifma
+- add movdiri, movdir64b, clwb, cldemote
+- WAITPKG instructions (tpause, umonitor, umwait) are supported.
+- MmapAllocator supports memfd with user-defined strings. see sample/memfd.cpp
+- strictly check address offset disp32 in a signed 32-bit integer. e.g., `ptr[(void*)0xffffffff]` causes an error.
+  - define `XBYAK_OLD_DISP_CHECK` if you need an old check, but the option will be remoevd.
+- add `jmp(mem, T_FAR)`, `call(mem, T_FAR)` `retf()` for far absolute indirect jump.
+- vnni instructions such as vpdpbusd supports vex encoding.
+- (break backward compatibility) `push(byte, imm)` (resp. `push(word, imm)`) forces to cast `imm` to 8(resp. 16) bit.
+- (Windows) `#include <winsock2.h>` has been removed from xbyak.h, so add it explicitly if you need it.
+- support exception-less mode see. [Exception-less mode](#exception-less-mode)
+- `XBYAK_USE_MMAP_ALLOCATOR` will be defined on Linux/macOS unless `XBYAK_DONT_USE_MMAP_ALLOCATOR` is defined.
 
-      jmp ("L2");
-      ...
-      a few mnemonics(8-bit displacement jmp)
-      ...
-    L("L2");
-
-      jmp ("L3", T_NEAR);
-      ...
-      a lot of mnemonics(32-bit displacement jmp)
-      ...
-    L("L3");
-
->Call hasUndefinedLabel() to verify your code has no undefined label.
-> you can use a label for immediate value of mov like as mov (eax, "L2");
-
-#### 1. support @@, @f, @b like MASM
-
-    L("@@"); // <A>
-      jmp("@b"); // jmp to <A>
-      jmp("@f"); // jmp to <B>
-    L("@@"); // <B>
-      jmp("@b"); // jmp to <B>
-      mov(eax, "@b");
-      jmp(eax); // jmp to <B>
-
-#### 2. localization of label by calling inLocalLabel(), outLocallabel().
-
-labels begining of period between inLocalLabel() and outLocalLabel()
-are dealed with local label.
-inLocalLabel() and outLocalLabel() can be nested.
-
-    void func1()
-    {
-        inLocalLabel();
-      L(".lp"); // <A> ; local label
-        ...
-        jmp(".lp"); // jmpt to <A>
-      L("aaa"); // global label
-        outLocalLabel();
-    }
-
-    void func2()
-    {
-        inLocalLabel();
-      L(".lp"); // <B> ; local label
-        func1();
-        jmp(".lp"); // jmp to <B>
-        inLocalLabel();
-    }
-
-### Label class
-
-L() and jxx() functions support a new Label class.
-
-      Label label1, label2;
-    L(label1);
-      ...
-      jmp(label1);
-      ...
-      jmp(label2);
-      ...
-    L(label2);
-
-Moreover, assignL(dstLabel, srcLabel) method binds dstLabel with srcLabel.
-
-      Label label1, label2;
-    L(label1);
-      ...
-      jmp(label2);
-      ...
-      assignL(label2, label1); // label2 <= label1
-
-The above jmp opecode jumps label1.
-
-* Restriction:
-* srcLabel must be used in L().
-* dstLabel must not be used in L().
-
-Label::getAddress() returns the address specified by the label instance and 0 if not specified.
-```
-// not AutoGrow mode
-Label  label;
-assert(label.getAddress() == 0);
-L(label);
-assert(label.getAddress() == getCurr());
-```
-
-### Rip
-```
-Label label;
-mov(eax, ptr [rip + label]); // eax = 4
-...
-
-L(label);
-dd(4);
-```
-```
-int x;
-...
-  mov(eax, ptr[rip + &x]); // throw exception if the difference between &x and current position is larger than 2GiB
-```
-### Code size
-The default max code size is 4096 bytes. Please set it in constructor of CodeGenerator() if you want to use large size.
-
-    class Quantize : public Xbyak::CodeGenerator {
-    public:
-      Quantize()
-        : CodeGenerator(8192)
-      {
-      }
-      ...
-    };
-
-### use user allocated memory
-
-You can make jit code on prepaired memory.
-
-    class Sample : public Xbyak::CodeGenerator {
-    public:
-        Sample(void *userPtr, size_t size)
-            : Xbyak::CodeGenerator(size, userPtr)
-        {
-            ...
-        }
-    };
-
-    const size_t codeSize = 1024;
-    uint8 buf[codeSize + 16];
-
-    // get 16-byte aligned address
-    uint8 *p = Xbyak::CodeArray::getAlignedAddress(buf);
-
-    // append executable attribute to the memory
-    Xbyak::CodeArray::protect(p, codeSize, true);
-
-    // construct your jit code on the memory
-    Sample s(p, codeSize);
-
->See *sample/test0.cpp*
+### Supported OS
 
-AutoGrow
--------------
+- Windows (Xp, Vista, 7, 10, 11) (32 / 64 bit)
+- Linux (32 / 64 bit)
+- macOS (Intel CPU)
 
-Under `AutoGrow` mode, Xbyak extends memory automatically if necessary.
-Call ready() before calling getCode() to calc address of jmp.
-```
-    struct Code : Xbyak::CodeGenerator {
-      Code()
-        : Xbyak::CodeGenerator(<default memory size>, Xbyak::AutoGrow)
-      {
-         ...
-      }
-    };
-    Code c;
-    c.ready(); // Don't forget to call this function
-```
->Don't use the address returned by getCurr() before calling ready().
->It may be invalid address.
->RESTRICTION : rip addressing is not supported in AutoGrow
+### Supported Compilers
 
-Macro
--------------
+Almost C++03 or later compilers for x86/x64 such as Visual Studio, g++, clang++, Intel C++ compiler and g++ on mingw/cygwin.
 
-* **XBYAK32** is defined on 32bit.
-* **XBYAK64** is defined on 64bit.
-* **XBYAK64_WIN** is defined on 64bit Windows(VC)
-* **XBYAK64_GCC** is defined on 64bit gcc, cygwin
-* define **XBYAK_NO_OP_NAMES** on gcc without `-fno-operator-names`
-* define **XBYAK_ENABLE_OMITTED_OPERAND** if you use omitted destination such as `vaddps(xmm2, xmm3);`(duplicated in the future)
+## License
 
-Sample
--------------
+[BSD-3-Clause License](http://opensource.org/licenses/BSD-3-Clause)
 
-* test0.cpp ; tiny sample of Xbyak(x86, x64)
-* quantize.cpp ; JIT optimized quantization by fast division(x86 only)
-* calc.cpp ; assemble and estimate a given polynomial(x86, x64)
-* bf.cpp ; JIT brainfuck(x86, x64)
+## Author
 
-License
--------------
+#### 光成滋生 Mitsunari Shigeo
+ [GitHub](https://github.com/herumi) | [Website (Japanese)](http://herumi.in.coocan.jp/) | [herumi@nifty.com](mailto:herumi@nifty.com)
 
-modified new BSD License
-http://opensource.org/licenses/BSD-3-Clause
+## Sponsors welcome
+[GitHub Sponsor](https://github.com/sponsors/herumi)
 
-The files under test/cybozu/ are copied from cybozulib(https://github.com/herumi/cybozulib/),
-which is licensed by BSD-3-Clause and are used for only tests.
-The header files under xbyak/ are independent of cybozulib.
+<!----------------------------------------------------------------------------->
 
-History
--------------
-* 2017/Jan/26 ver 5.41 add prefetchwt1 and support for scale == 0(thanks to rsdubtso)
-* 2016/Dec/14 ver 5.40 add Label::getAddress() method to get the pointer specified by the label
-* 2016/Dec/09 ver 5.34 fix handling of negative offsets when encoding disp8N(thanks to rsdubtso)
-* 2016/Dec/08 ver 5.33 fix encoding of vpbroadcast{b,w,d,q}, vpinsr{b,w}, vpextr{b,w} for disp8N
-* 2016/Dec/01 ver 5.32 rename __xgetbv() to _xgetbv() to support clang for Visual Studio(thanks to freiro)
-* 2016/Nov/27 ver 5.31 rename AVX512_4VNNI to AVX512_4VNNIW
-* 2016/Nov/27 ver 5.30 add AVX512_4VNNI, AVX512_4FMAPS instructions(thanks to rsdubtso)
-* 2016/Nov/26 ver 5.20 add detection of AVX512_4VNNI and AVX512_4FMAPS(thanks to rsdubtso)
-* 2016/Nov/20 ver 5.11 lost vptest for ymm(thanks to gregory38)
-* 2016/Nov/20 ver 5.10 add addressing [rip+&var]
-* 2016/Sep/29 ver 5.03 fix detection ERR_INVALID_OPMASK_WITH_MEMORY(thanks to PVS-Studio)
-* 2016/Aug/15 ver 5.02 xbyak does not include xbyak_bin2hex.h
-* 2016/Aug/15 ver 5.011 fix detection of version of gcc 5.4
-* 2016/Aug/03 ver 5.01 disable omitted operand
-* 2016/Jun/24 ver 5.00 support avx-512 instruction set
-* 2016/Jun/13 avx-512 add mask instructions
-* 2016/May/05 ver 4.91 add detection of AVX-512 to Xbyak::util::Cpu
-* 2016/Mar/14 ver 4.901 comment to ready() function(thanks to skmp)
-* 2016/Feb/04 ver 4.90 add jcc(const void *addr);
-* 2016/Jan/30 ver 4.89 vpblendvb supports ymm reg(thanks to John Funnell)
-* 2016/Jan/24 ver 4.88 lea, cmov supports 16-bit register(thanks to whyisthisfieldhere)
-* 2015/Oct/05 ver 4.87 support segment selectors
-* 2015/Aug/18 ver 4.86 fix [rip + label] addressing with immediate value(thanks to whyisthisfieldhere)
-* 2015/Aug/10 ver 4.85 Address::operator==() is not correct(thanks to inolen)
-* 2015/Jun/22 ver 4.84 call() support variadic template if available(thanks to randomstuff)
-* 2015/Jun/16 ver 4.83 support movbe(thanks to benvanik)
-* 2015/May/24 ver 4.82 support detection of F16C
-* 2015/Apr/25 ver 4.81 fix the condition to throw exception for setSize(thanks to whyisthisfieldhere)
-* 2015/Apr/22 ver 4.80 rip supports label(thanks to whyisthisfieldhere)
-* 2015/Jar/28 ver 4.71 support adcx, adox, cmpxchg, rdseed, stac
-* 2014/Oct/14 ver 4.70 support MmapAllocator
-* 2014/Jun/13 ver 4.62 disable warning of VC2014
-* 2014/May/30 ver 4.61 support bt, bts, btr, btc
-* 2014/May/28 ver 4.60 support vcvtph2ps, vcvtps2ph
-* 2014/Apr/11 ver 4.52 add detection of rdrand
-* 2014/Mar/25 ver 4.51 remove state information of unreferenced labels
-* 2014/Mar/16 ver 4.50 support new Label
-* 2014/Mar/05 ver 4.40 fix wrong detection of BMI/enhanced rep on VirtualBox
-* 2013/Dec/03 ver 4.30 support Reg::cvt8(), cvt16(), cvt32(), cvt64()
-* 2013/Oct/16 ver 4.21 label support std::string
-* 2013/Jul/30 ver 4.20 [break backward compatibility] split Reg32e class into RegExp(base+index*scale+disp) and Reg32e(means Reg32 or Reg64)
-* 2013/Jul/04 ver 4.10 [break backward compatibility] change the type of Xbyak::Error from enum to a class
-* 2013/Jun/21 ver 4.02 add putL(LABEL) function to put the address of the label
-* 2013/Jun/21 ver 4.01 vpsllw, vpslld, vpsllq, vpsraw, vpsrad, vpsrlw, vpsrld, vpsrlq support (ymm, ymm, xmm).
-                       support vpbroadcastb, vpbroadcastw, vpbroadcastd, vpbroadcastq(thanks to Gabest).
-* 2013/May/30 ver 4.00 support AVX2, VEX-encoded GPR-instructions
-* 2013/Mar/27 ver 3.80 support mov(reg, "label");
-* 2013/Mar/13 ver 3.76 add cqo(), jcxz(), jecxz(), jrcxz()
-* 2013/Jan/15 ver 3.75 add setSize() to modify generated code
-* 2013/Jan/12 ver 3.74 add CodeGenerator::reset() ; add Allocator::useProtect()
-* 2013/Jan/06 ver 3.73 use unordered_map if possible
-* 2012/Dec/04 ver 3.72 eax, ebx, ... are member variables of CodeGenerator(revert), Xbyak::util::eax, ... are static const.
-* 2012/Nov/17 ver 3.71 and_(), or_(), xor_(), not_() are available if XBYAK_NO_OP_NAMES is not defined.
-* 2012/Nov/17 change eax, ebx, ptr and so on in CodeGenerator as static member and alias of them are defined in Xbyak::util.
-* 2012/Nov/09 ver 3.70 XBYAK_NO_OP_NAMES macro is added to use and_() instead of and() (thanks to Mattias)
-* 2012/Nov/01 ver 3.62 add fwait/fnwait/finit/fninit
-* 2012/Nov/01 ver 3.61 add fldcw/fstcw
-* 2012/May/03 ver 3.60 change interface of Allocator
-* 2012/Mar/23 ver 3.51 fix userPtr mode
-* 2012/Mar/19 ver 3.50 support AutoGrow mode
-* 2011/Nov/09 ver 3.05 fix bit property of rip addresing / support movsxd
-* 2011/Aug/15 ver 3.04 fix dealing with imm8 such as add(dword [ebp-8], 0xda); (thanks to lolcat)
-* 2011/Jun/16 ver 3.03 fix __GNUC_PREREQ macro for Mac gcc(thanks to t_teruya)
-* 2011/Apr/28 ver 3.02 do not use xgetbv on Mac gcc
-* 2011/May/24 ver 3.01 fix typo of OSXSAVE
-* 2011/May/23 ver 3.00 add vcmpeqps and so on
-* 2011/Feb/16 ver 2.994 beta add vmovq for 32-bit mode(I forgot it)
-* 2011/Feb/16 ver 2.993 beta remove cvtReg to avoid thread unsafe
-* 2011/Feb/10 ver 2.992 beta support one argument syntax for fadd like nasm
-* 2011/Feb/07 ver 2.991 beta fix pextrw reg, xmm, imm(Thanks to Gabest)
-* 2011/Feb/04 ver 2.99 beta support AVX
-* 2010/Dec/08 ver 2.31 fix ptr [rip + 32bit offset], support rdtscp
-* 2010/Oct/19 ver 2.30 support pclmulqdq, aesdec, aesdeclast, aesenc, aesenclast, aesimc, aeskeygenassist
-* 2010/Jun/07 ver 2.29 fix call(<label>)
-* 2010/Jun/17 ver 2.28 move some member functions to public
-* 2010/Jun/01 ver 2.27 support encoding of mov(reg64, imm) like yasm(not nasm)
-* 2010/May/24 ver 2.26 fix sub(rsp, 1000)
-* 2010/Apr/26 ver 2.25 add jc/jnc(I forgot to implement them...)
-* 2010/Apr/16 ver 2.24 change the prototype of rewrite() method
-* 2010/Apr/15 ver 2.23 fix align() and xbyak_util.h for Mac
-* 2010/Feb/16 ver 2.22 fix inLocalLabel()/outLocalLabel()
-* 2009/Dec/09 ver 2.21 support cygwin(gcc 4.3.2)
-* 2009/Nov/28 support a part of FPU
-* 2009/Jun/25 fix mov(qword[rax], imm); (thanks to Martin)
-* 2009/Mar/10 fix redundant REX.W prefix on jmp/call reg64
-* 2009/Feb/24 add movq reg64, mmx/xmm; movq mmx/xmm, reg64
-* 2009/Feb/13 movd(xmm7, dword[eax]) drops 0x66 prefix (thanks to Gabest)
-* 2008/Dec/30 fix call in short relative address(thanks to kato san)
-* 2008/Sep/18 support @@, @f, @b and localization of label(thanks to nobu-q san)
-* 2008/Sep/18 support (ptr[rip + 32bit offset]) (thanks to Dango-Chu san)
-* 2008/Jun/03 fix align(). mov(ptr[eax],1) throws ERR_MEM_SIZE_IS_NOT_SPECIFIED.
-* 2008/Jun/02 support memory interface allocated by user
-* 2008/May/26 fix protect() to avoid invalid setting(thanks to shinichiro_h san)
-* 2008/Apr/30 add cmpxchg16b, cdqe
-* 2008/Apr/29 support x64
-* 2008/Apr/14 code refactoring
-* 2008/Mar/12 add bsr/bsf
-* 2008/Feb/14 fix output of sub eax, 1234 (thanks to Robert)
-* 2007/Nov/5  support lock, xadd, xchg
-* 2007/Nov/2  support SSSE3/SSE4 (thanks to Dango-Chu san)
-* 2007/Feb/4  fix the bug that exception doesn't occur under the condition which the offset of jmp mnemonic without T_NEAR is over 127.
-* 2007/Jan/21 fix the bug to create address like [disp] select smaller representation for mov (eax|ax|al, [disp])
-* 2007/Jan/4  first version
+[Badge Build]: https://github.com/herumi/xbyak/actions/workflows/main.yml/badge.svg
+[Build Status]: https://github.com/herumi/xbyak/actions/workflows/main.yml
 
-Author
--------------
+[License]: COPYRIGHT
 
-MITSUNARI Shigeo(herumi@nifty.com)
+[Changelog]: doc/changelog.md
+[Install]: doc/install.md
+[Usage]: doc/usage.md
 
diff --git a/dep/xbyak/readme.txt b/dep/xbyak/readme.txt
index 4ab1582e4..975539154 100644
--- a/dep/xbyak/readme.txt
+++ b/dep/xbyak/readme.txt
@@ -1,5 +1,5 @@
 
-    C++用x86(IA-32), x64(AMD64, x86-64) JITアセンブラ Xbyak 5.41
+    C++用x86(IA-32), x64(AMD64, x86-64) JITアセンブラ Xbyak 6.73
 
 -----------------------------------------------------------------------------
 ◎概要
@@ -22,21 +22,21 @@
     Intel Mac
     などで動作確認をしています。
 
-※ Xbyakはデフォルトでand(), or(), xor(), not()関数を使います。
-gccではそれらを演算子として解釈してしまうため、-fno-operator-namesオプションを追加してコンパイルしてください。
-あるいはXBYAK_NO_OP_NAMESを定義してand_(), or_(), xor_(), not_()を使ってください。
-and_(), or_(), xor_(), not_()はXBYAK_NO_OP_NAMESされていないときでも使えます。
+※ and, orなどの代わりにand_, or_を使用してください。
+and, orなどを使いたい場合は-fno-operator-namesをgcc/clangに指定してください。
 
 -----------------------------------------------------------------------------
 ◎準備
 xbyak.h
 xbyak_bin2hex.h
-xbyak_mnemonic.h
 これらを同一のパスに入れてインクルードパスに追加してください。
 
 Linuxではmake installで/usr/local/include/xbyakにコピーされます。
 -----------------------------------------------------------------------------
 ◎下位互換性の破れ
+* push byte, immまたはpush word, immが下位8bit, 16bitにキャストした値を使うように変更。
+* (Windows) `<winsock2.h>`をincludeしなくなったので必要なら明示的にincludeしてください。
+* XBYAK_USE_MMAP_ALLOCATORがデフォルトで有効になりました。従来の方式にする場合はXBYAK_DONT_USE_MMAP_ALLOCATORを定義してください。
 * Xbyak::Errorの型をenumからclassに変更
 ** 従来のenumの値をとるにはintにキャストしてください。
 * (古い)Reg32eクラスを(新しい)Reg32eとRegExpに分ける。
@@ -46,6 +46,13 @@ Linuxではmake installで/usr/local/include/xbyakにコピーされます。
 -----------------------------------------------------------------------------
 ◎新機能
 
+例外なしモード追加
+XBYAK_NO_EXCEPTIONを定義してコンパイルするとgcc/clangで-fno-exceptionsオプションでコンパイルできます。
+エラーは例外の代わりに`Xbyak::GetError()`で通達されます。
+この値が0でなければ何か問題が発生しています。
+この値は自動的に変更されないので`Xbyak::ClearError()`でリセットしてください。
+`CodeGenerator::reset()`は`ClearError()`を呼びます。
+
 MmapAllocator追加
 これはUnix系OSでのみの仕様です。XBYAK_USE_MMAP_ALLOCATORを使うと利用できます。
 デフォルトのAllocatorはメモリ確保時にposix_memalignを使います。
@@ -54,7 +61,6 @@ map countの最大値は/proc/sys/vm/max_map_countに書かれています。
 デフォルトでは3万個ほどのXbyak::CodeGeneratorインスタンスを生成するとエラーになります。
 test/mprotect_test.cppで確認できます。
 これを避けるためにはmmapを使うMmapAllocatorを使ってください。
-将来この挙動がデフォルトになるかもしれません。
 
 
 AutoGrowモード追加
@@ -155,15 +161,20 @@ vcvtpd2dq xmm19, [eax+32]{1to4}         --> vcvtpd2dq(xmm19, yword_b [eax+32]);
 
 vfpclassps k5{k3}, zword [rax+64], 5    --> vfpclassps(k5|k3, zword [rax+64], 5); // specify m512
 vfpclasspd k5{k3}, [rax+64]{1to2}, 5    --> vfpclasspd(k5|k3, xword_b [rax+64], 5); // broadcast 64-bit to 128-bit
-vfpclassps k5{k3}, [rax+64]{1to4}, 5    --> vfpclassps(k5|k3, xword_b [rax+64], 5); // broadcast 32-bit to 128-bit
-
+vfpclassps k5{k3}, [rax+64]{1to4}, 5    --> vfpclassps(k5|k3, xword_b [rax+64], 5); // broadcast 64-bit to 256-bit
 
+vpdpbusd(xm0, xm1, xm2); // default encoding is EVEX
+vpdpbusd(xm0, xm1, xm2, EvexEncoding); // same as the above
+vpdpbusd(xm0, xm1, xm2, VexEncoding); // VEX encoding
+setDefaultEncoding(VexEncoding); // default encoding is VEX
+vpdpbusd(xm0, xm1, xm2); // VEX encoding
 注意
 * k1, ..., k7 は新しいopmaskレジスタです。
 * z, sae, rn-sae, rd-sae, ru-sae, rz-saeの代わりにT_z, T_sae, T_rn_sae, T_rd_sae, T_ru_sae, T_rz_saeを使ってください。
 * `k4 | k3`と`k3 | k4`は意味が異なります。
 * {1toX}の代わりにptr_bを使ってください。Xは自動的に決まります。
 * 一部の命令はメモリサイズを指定するためにxword/yword/zword(_b)を使ってください。
+* setDefaultEncoding()でencoding省略時のEVEX/VEXを設定できます。
 
 ・ラベル
 
@@ -245,8 +256,8 @@ void func2()
 
 更にラベルの割り当てを行うassignL(dstLabel, srcLabel)という命令も追加されました。
 
-      Label label1, label2;
-    L(label1);
+      Label label2;
+    Label label1 = L(); // Label label1; L(label1);と同じ意味
       ...
       jmp(label2);
       ...
@@ -268,6 +279,24 @@ L(label);
 assert(label.getAddress(), getCurr());
 ```
 
+4. farジャンプ
+
+`jmp(mem, T_FAR)`, `call(mem, T_FAR)`, `retf()`をサポートします。
+サイズを明示するために`ptr`の代わりに`word|dword|qword`を利用してください。
+
+32bit
+```
+jmp(word[eax], T_FAR);  // jmp m16:16(FF /5)
+jmp(dword[eax], T_FAR); // jmp m16:32(FF /5)
+```
+
+64bit
+```
+jmp(word[rax], T_FAR);  // jmp m16:16(FF /5)
+jmp(dword[rax], T_FAR); // jmp m16:32(FF /5)
+jmp(qword[rax], T_FAR); // jmp m16:64(REX.W FF /5)
+```
+
 ・Xbyak::CodeGenerator()コンストラクタインタフェース
 
 @param maxSize [in] コード生成最大サイズ(デフォルト4096byte)
@@ -309,6 +338,41 @@ bool CodeArray::protect(const void *addr, size_t size, bool canExec);
 */
 uint8 *CodeArray::getAlignedAddress(uint8 *addr, size_t alignedSize = ALIGN_SIZE);
 
+・read/execモード
+デフォルトのCodeGeneratorはコンストラクト時にJIT用の領域をread/write/execモードに設定して利用します。
+コード生成時はread/writeでコード実行時にはread/execにしたい場合、次のようにしてください。
+
+struct Code : Xbyak::CodeGenerator {
+    Code()
+        : Xbyak::CodeGenerator(4096, Xbyak::DontUseProtect) // JIT領域をread/writeのままコード生成
+    {
+        mov(eax, 123);
+        ret();
+    }
+};
+
+Code c;
+c.setProtectModeRE(); // read/execモードに変更
+// JIT領域を実行
+
+AutoGrowの場合はreadyの代わりにreadyRE()を読んでください。
+
+struct Code : Xbyak::CodeGenerator {
+    Code()
+        : Xbyak::CodeGenerator(4096, Xbyak::AutoGrow) // JIT領域をread/writeのままコード生成
+    {
+        mov(eax, 123);
+        ret();
+    }
+};
+
+Code c;
+c.readyRE(); // read/exeモードに変更
+// JIT領域を実行
+
+setProtectModeRW()を呼ぶと領域が元のread/execモードに戻ります。
+
+
 その他詳細は各種サンプルを参照してください。
 -----------------------------------------------------------------------------
 ◎マクロ
@@ -335,14 +399,104 @@ http://opensource.org/licenses/BSD-3-Clause
 sample/{echo,hello}.bfは http://www.kmonos.net/alang/etc/brainfuck.php から
 いただきました。
 
-test/cybozu/以下のファイルはcybozulib(https://github.com/herumi/cybozulib/)
-の一部を使っています。cybozulibはBSD-3-Clauseライセンスです。
-cybozulibは単体テストでのみ利用されていて、xbyak/ディレクトリ以下のヘッダ
-ファイルはcybozulibとは独立に利用できます。
-
 -----------------------------------------------------------------------------
 ◎履歴
 
+2023/08/07 ver 6.73 sha512/sm3/sm4/avx-vnni-int16追加
+2023/08/02 ver 6.72 xabort, xbegin, xend追加
+2023/07/27 ver 6.71 Allocatorでhuge pageを考慮する。
+2023/07/05 ver 6.70 vpclmulqdqのailas追加
+2023/06/27 ver 6.69.2 `TypeT operator|`にconstexpr追加(thanks to Wunkolo)
+2023/03/23 ver 6.69.1 xsave判定追加(thanks to Wunkolo)
+2023/02/20 ver 6.69 util::CpuがAMD対応 UINTR命令対応
+2022/12/07 ver 6.68 prefetchit{0,1}サポート
+2022/11/30 ver 6.67 CMPccXADDサポート
+2022/11/25 ver 6.66 RAO-INTサポート
+2022/11/22 ver 6.65 x32動作確認
+2022/11/04 ver 6.64 vmov*命令をmaskつきアドレッシング対応修正
+2022/10/06 ver 6.63 AVX-IFMA用のvpmadd52{h,l}uq対応
+2022/10/05          amx_fp16/avx_vnni_int8/avx_ne_convertt対応とsetDefaultEncoding()追加
+2022/09/15 ver 6.62 serialize追加
+2022/08/02 ver 6.61.1 noexceptはVisual Studio 2015以降対応
+2022/07/29 ver 6.61 movzx eax, ahがエラーになるのを修正
+2022/06/16 ver 6.60.2 GFNI, VAES, VPCLMULQDQの判定修正
+2022/06/15 ver 6.60.1 Visual Studio /O0でXbyak::util::Cpuがリンクエラーになるのに対応
+2022/06/06 ver 6.60 バージョンのつけ方を数値が戻らないように変更
+2022/06/01 ver 6.06 Cpu::TypeクラスのリファクタリングとXBYAK_USE_MEMFDが定義されたときのMmapAllocatorの改善
+2022/05/20 ver 6.052 Cpu::operator==()を正しく定義
+2022/05/13 ver 6.051 XYBAK_NO_EXCEPTIONを定義したときのCpuクラスのコンパイルエラー修正
+2022/05/12 ver 6.05 movdiri, movdir64b, clwb, cldemoteを追加
+2022/04/05 ver 6.04 tpause, umonitor, umwaitを追加
+2022/03/08 ver 6.03 MmapAllocatorがmemfd用のユーザ定義文字列をサポート
+2022/01/28 ver 6.02 dispacementの32bit範囲チェックの厳密化
+2021/12/14 ver 6.01 T_FAR jump/callとretfをサポート
+2021/09/14 ver 6.00 AVX512-FP16を完全サポート
+2021/09/09 ver 5.997 vrndscale*を{sae}をサポートするよう修正
+2021/09/03 ver 5.996 v{add,sub,mul,div,max,min}{sd,ss}をT_rd_saeなどをサポートするよう修正
+2021/08/15 ver 5.995 Linux上でXBYAK_USE_MEMFDが定義されたなら/proc/self/mapsにラベル追加
+2021/06/17 ver 5.994 マスクレジスタ用のvcmpXX{ps,pd,ss,sd}のalias追加
+2021/06/06 ver 5.993 gather/scatterのレジスタの組み合わせの厳密なチェック
+2021/05/09 ver 5.992 endbr32とendbr64のサポート
+2020/11/16 ver 5.991 g++-5のC++14でconstexpr機能の抑制
+2020/10/19 ver 5.99 VNNI命令サポート(Thanks to akharito)
+2020/10/17 ver 5.98 [scale * reg]のサポート
+2020/09/08 ver 5.97 uint32などをuint32_tに置換
+2020/08/28 ver 5.95 レジスタクラスのコンストラクタがconstexprに対応(C++14以降)
+2020/08/04 ver 5.941 `CodeGenerator::reset()`が`ClearError()`を呼ぶように変更
+2020/07/28 ver 5.94 #include <winsock2.h>の削除 (only windows)
+2020/07/21 ver 5.93 例外なしモード追加
+2020/06/30 ver 5.92 Intel AMX命令サポート (Thanks to nshustrov)
+2020/06/19 ver 5.913 32ビット環境でXBYAK64を定義したときのmov(r64, imm64)を修正
+2020/06/19 ver 5.912 macOSの古いXcodeでもMAP_JITを有効にする(Thanks to rsdubtso)
+2020/05/10 ver 5.911 Linux/macOSでXBYAK_USE_MMAP_ALLOCATORがデフォルト有効になる
+2020/04/20 ver 5.91 マスクレジスタk0を受け入れる(マスクをしない)
+2020/04/09 ver 5.90 kmov{b,w,d,q}がサポートされないレジスタを受けると例外を投げる
+2020/02/26 ver 5.891 zm0のtype修正
+2020/01/03 ver 5.89 vfpclasspdの処理エラー修正
+2019/12/20 ver 5.88 Windowsでのコンパイルエラー修正
+2019/12/19 ver 5.87 未定義ラベルへのjmp命令のデフォルト挙動をT_NEARにするsetDefaultJmpNEAR()を追加
+2019/12/13 ver 5.86 [変更] -fno-operator-namesが指定されたときは5.84以前の挙動に戻す
+2019/12/07 ver 5.85 mmapにMAP_JITフラグを追加(macOS mojave以上)
+2019/11/29 ver 5.84 [変更] XBYAK_USE_OP_NAMESが定義されていない限りXBYAK_NO_OP_NAMESが定義されるように変更
+2019/10/12 ver 5.83 exit(1)の除去
+2019/09/23 ver 5.82 monitorx, mwaitx, clzero対応 (thanks to MagurosanTeam)
+2019/09/14 ver 5.81 いくつかの一般命令をサポート
+2019/08/01 ver 5.802 AVX512_BF16判定修正 (thanks to vpirogov)
+2019/05/27 support vp2intersectd, vp2intersectq (not tested)
+2019/05/26 ver 5.80 support vcvtne2ps2bf16, vcvtneps2bf16, vdpbf16ps
+2019/04/27 ver 5.79 vcmppd/vcmppsのptr_b対応忘れ(thanks to jkopinsky)
+2019/04/15 ver 5.78 Reg::changeBit()のリファクタリング(thanks to MerryMage)
+2019/03/06 ver 5.77 LLCキャッシュを共有数CPU数の修整(by densamoilov)
+2019/01/17 ver 5.76 Cpu::getNumCores()追加(by shelleygoel)
+2018/10/31 ver 5.751 互換性のためにXbyak::CastToの復元
+2018/10/29 ver 5.75 LabelManagerのデストラクタでLabelから参照を切り離す
+2018/10/21 ver 5.74 RegRip +/intの形をサポート Xbyak::CastToを削除
+2018/10/15 util::StackFrameでmovの代わりにpush/popを使う
+2018/09/19 ver 5.73 vpslld, vpslldq, vpsllwなどの(reg, mem, imm8)に対するevexエンコーディング修整
+2018/09/19 ver 5.72 fix the encoding of vinsertps for disp8N(Thanks to petercaday)
+2018/08/27 ver 5.71 新しいlabelインスタンスを返すL()を追加
+2018/08/27 ver 5.70 read/exec設定のためのsetProtectMode()とDontUseProtectの追加
+2018/08/24 ver 5.68 indexが16以上のVSIBエンコーディングのバグ修正(thanks to petercaday)
+2018/08/14 ver 5.67 Addressクラス内のmutableを削除 ; fix setCacheHierarchy for cloud vm
+2018/07/26 ver 5.661 mingw64対応
+2018/07/24 ver 5.66 protect()のmodeにCodeArray::PROTECT_REを追加
+2018/06/26 ver 5.65 fix push(qword [mem])
+2018/03/07 ver 5.64 Cpu()の中でzero divisionが出ることがあるのを修正
+2018/02/14 ver 5.63 Cpu::setCacheHierarchy()の修正とclang<3.9のためのEvexModifierZero修正(thanks to mgouicem)
+2018/02/13 ver 5.62 Cpu::setCacheHierarchy() by mgouicem and rsdubtso
+2018/02/07 ver 5.61 vmov*がmem{k}{z}形式対応(忘れてた)
+2018/01/24 ver 5.601 xword, ywordなどをXbyak::util名前空間に追加
+2018/01/05 ver 5.60 Ice lake系命令対応(319433-030.pdf)
+2017/08/22 ver 5.53 mpxエンコーディングバグ修正, bnd()プレフィクス追加
+2017/08/18 ver 5.52 align修正(thanks to MerryMage)
+2017/08/17 ver 5.51 multi-byte nop追加 align()はそれを使用する(thanks to inolen)
+2017/08/08 ver 5.50 mpx追加(thanks to magurosan)
+2017/08/08 ver 5.45 sha追加(thanks to magurosan)
+2017/08/08 ver 5.44 prefetchw追加(thanks to rsdubtso)
+2017/07/12 ver 5.432 PVS-studioの警告を減らす
+2017/07/09 ver 5.431 hasRex()修正 (影響なし) (thanks to drillsar)
+2017/05/14 ver 5.43 CodeGenerator::resetSize()修正(thanks to gibbed)
+2017/05/13 ver 5.42 movs{b,w,d,q}追加
 2017/01/26 ver 5.41 prefetcwt1追加とscale == 0対応(thanks to rsdubtso)
 2016/12/14 ver 5.40 Labelが示すアドレスを取得するLabel::getAddress()追加
 2016/12/07 ver 5.34 disp8N時の負のオフセット処理の修正(thanks to rsdubtso)
@@ -458,7 +612,3 @@ cybozulibは単体テストでのみ利用されていて、xbyak/ディレク
 ◎著作権者
 
 光成滋生(MITSUNARI Shigeo, herumi@nifty.com)
-
----
-$Revision: 1.56 $
-$Date: 2010/04/16 11:58:22 $
diff --git a/dep/xbyak/xbyak/xbyak.h b/dep/xbyak/xbyak/xbyak.h
index 431dc4907..01fddf750 100644
--- a/dep/xbyak/xbyak/xbyak.h
+++ b/dep/xbyak/xbyak/xbyak.h
@@ -9,10 +9,8 @@
 	@note modified new BSD license
 	http://opensource.org/licenses/BSD-3-Clause
 */
-#ifndef XBYAK_NO_OP_NAMES
-	#if not +0 // trick to detect whether 'not' is operator or not
-		#error "use -fno-operator-names option if you want to use and(), or(), xor(), not() as function names, Or define XBYAK_NO_OP_NAMES and use and_(), or_(), xor_(), not_()."
-	#endif
+#if (not +0) && !defined(XBYAK_NO_OP_NAMES) // trick to detect whether 'not' is operator or not
+	#define XBYAK_NO_OP_NAMES
 #endif
 
 #include <stdio.h> // for debug print
@@ -26,7 +24,9 @@
 
 // #define XBYAK_DISABLE_AVX512
 
-//#define XBYAK_USE_MMAP_ALLOCATOR
+#if !defined(XBYAK_USE_MMAP_ALLOCATOR) && !defined(XBYAK_DONT_USE_MMAP_ALLOCATOR)
+	#define XBYAK_USE_MMAP_ALLOCATOR
+#endif
 #if !defined(__GNUC__) || defined(__MINGW32__)
 	#undef XBYAK_USE_MMAP_ALLOCATOR
 #endif
@@ -40,6 +40,8 @@
 // This covers -std=(gnu|c)++(0x|11|1y), -stdlib=libc++, and modern Microsoft.
 #if ((defined(_MSC_VER) && (_MSC_VER >= 1600)) || defined(_LIBCPP_VERSION) ||\
 	 			 ((__cplusplus >= 201103) || defined(__GXX_EXPERIMENTAL_CXX0X__)))
+	#include <unordered_set>
+	#define XBYAK_STD_UNORDERED_SET std::unordered_set
 	#include <unordered_map>
 	#define XBYAK_STD_UNORDERED_MAP std::unordered_map
 	#define XBYAK_STD_UNORDERED_MULTIMAP std::unordered_multimap
@@ -49,33 +51,60 @@
 	libstdcxx 20070719 (from GCC 4.2.1, the last GPL 2 version).
 */
 #elif XBYAK_GNUC_PREREQ(4, 5) || (XBYAK_GNUC_PREREQ(4, 2) && __GLIBCXX__ >= 20070719) || defined(__INTEL_COMPILER) || defined(__llvm__)
+	#include <tr1/unordered_set>
+	#define XBYAK_STD_UNORDERED_SET std::tr1::unordered_set
 	#include <tr1/unordered_map>
 	#define XBYAK_STD_UNORDERED_MAP std::tr1::unordered_map
 	#define XBYAK_STD_UNORDERED_MULTIMAP std::tr1::unordered_multimap
 
 #elif defined(_MSC_VER) && (_MSC_VER >= 1500) && (_MSC_VER < 1600)
+	#include <unordered_set>
+	#define XBYAK_STD_UNORDERED_SET std::tr1::unordered_set
 	#include <unordered_map>
 	#define XBYAK_STD_UNORDERED_MAP std::tr1::unordered_map
 	#define XBYAK_STD_UNORDERED_MULTIMAP std::tr1::unordered_multimap
 
 #else
+	#include <set>
+	#define XBYAK_STD_UNORDERED_SET std::set
 	#include <map>
 	#define XBYAK_STD_UNORDERED_MAP std::map
 	#define XBYAK_STD_UNORDERED_MULTIMAP std::multimap
 #endif
 #ifdef _WIN32
-	#include <winsock2.h>
+	#ifndef WIN32_LEAN_AND_MEAN
+		#define WIN32_LEAN_AND_MEAN
+	#endif
 	#include <windows.h>
 	#include <malloc.h>
+	#ifdef _MSC_VER
+		#define XBYAK_TLS __declspec(thread)
+	#else
+		#define XBYAK_TLS __thread
+	#endif
 #elif defined(__GNUC__)
 	#include <unistd.h>
 	#include <sys/mman.h>
 	#include <stdlib.h>
+	#define XBYAK_TLS __thread
+#endif
+#if defined(__APPLE__) && !defined(XBYAK_DONT_USE_MAP_JIT)
+	#define XBYAK_USE_MAP_JIT
+	#include <sys/sysctl.h>
+	#ifndef MAP_JIT
+		#define MAP_JIT 0x800
+	#endif
 #endif
 #if !defined(_MSC_VER) || (_MSC_VER >= 1600)
 	#include <stdint.h>
 #endif
 
+// MFD_CLOEXEC defined only linux 3.17 or later.
+// Android wraps the memfd_create syscall from API version 30.
+#if !defined(MFD_CLOEXEC) || (defined(__ANDROID__) && __ANDROID_API__ < 30)
+	#undef XBYAK_USE_MEMFD
+#endif
+
 #if defined(_WIN64) || defined(__MINGW64__) || (defined(__CYGWIN__) && defined(__x86_64__))
 	#define XBYAK64_WIN
 #elif defined(__x86_64__)
@@ -89,8 +118,22 @@
 	#endif
 #endif
 
-#if (__cplusplus >= 201103) || (_MSC_VER >= 1800)
+#if (__cplusplus >= 201103) || (defined(_MSC_VER) && _MSC_VER >= 1900)
+	#undef XBYAK_TLS
+	#define XBYAK_TLS thread_local
 	#define XBYAK_VARIADIC_TEMPLATE
+	#define XBYAK_NOEXCEPT noexcept
+#else
+	#define XBYAK_NOEXCEPT throw()
+#endif
+
+// require c++14 or later
+// Visual Studio 2017 version 15.0 or later
+// g++-6 or later
+#if ((__cplusplus >= 201402L) && !(!defined(__clang__) && defined(__GNUC__) && (__GNUC__ <= 5))) || (defined(_MSC_VER) && _MSC_VER >= 1910)
+	#define XBYAK_CONSTEXPR constexpr
+#else
+	#define XBYAK_CONSTEXPR
 #endif
 
 #ifdef _MSC_VER
@@ -101,25 +144,28 @@
 	#pragma warning(disable : 4127) /* constant expresison */
 #endif
 
+// disable -Warray-bounds because it may be a bug of gcc. https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104603
+#if defined(__GNUC__) && !defined(__clang__)
+	#define XBYAK_DISABLE_WARNING_ARRAY_BOUNDS
+	#pragma GCC diagnostic push
+	#pragma GCC diagnostic ignored "-Warray-bounds"
+#endif
+
 namespace Xbyak {
 
 enum {
 	DEFAULT_MAX_CODE_SIZE = 4096,
-	VERSION = 0x5410 /* 0xABCD = A.BC(D) */
+	VERSION = 0x6730 /* 0xABCD = A.BC(.D) */
 };
 
 #ifndef MIE_INTEGER_TYPE_DEFINED
 #define MIE_INTEGER_TYPE_DEFINED
-#ifdef _MSC_VER
-	typedef unsigned __int64 uint64;
-	typedef __int64 sint64;
-#else
-	typedef uint64_t uint64;
-	typedef int64_t sint64;
-#endif
-typedef unsigned int uint32;
-typedef unsigned short uint16;
-typedef unsigned char uint8;
+// for backward compatibility
+typedef uint64_t uint64;
+typedef int64_t sint64;
+typedef uint32_t uint32;
+typedef uint16_t uint16;
+typedef uint8_t uint8;
 #endif
 
 #ifndef MIE_ALIGN
@@ -177,78 +223,123 @@ enum {
 	ERR_INVALID_OPMASK_WITH_MEMORY,
 	ERR_INVALID_ZERO,
 	ERR_INVALID_RIP_IN_AUTO_GROW,
-	ERR_INTERNAL
+	ERR_INVALID_MIB_ADDRESS,
+	ERR_X2APIC_IS_NOT_SUPPORTED,
+	ERR_NOT_SUPPORTED,
+	ERR_SAME_REGS_ARE_INVALID,
+	ERR_INTERNAL // Put it at last.
 };
 
+inline const char *ConvertErrorToString(int err)
+{
+	static const char *errTbl[] = {
+		"none",
+		"bad addressing",
+		"code is too big",
+		"bad scale",
+		"esp can't be index",
+		"bad combination",
+		"bad size of register",
+		"imm is too big",
+		"bad align",
+		"label is redefined",
+		"label is too far",
+		"label is not found",
+		"code is not copyable",
+		"bad parameter",
+		"can't protect",
+		"can't use 64bit disp(use (void*))",
+		"offset is too big",
+		"MEM size is not specified",
+		"bad mem size",
+		"bad st combination",
+		"over local label",
+		"under local label",
+		"can't alloc",
+		"T_SHORT is not supported in AutoGrow",
+		"bad protect mode",
+		"bad pNum",
+		"bad tNum",
+		"bad vsib addressing",
+		"can't convert",
+		"label is not set by L()",
+		"label is already set by L()",
+		"bad label string",
+		"err munmap",
+		"opmask is already set",
+		"rounding is already set",
+		"k0 is invalid",
+		"evex is invalid",
+		"sae(suppress all exceptions) is invalid",
+		"er(embedded rounding) is invalid",
+		"invalid broadcast",
+		"invalid opmask with memory",
+		"invalid zero",
+		"invalid rip in AutoGrow",
+		"invalid mib address",
+		"x2APIC is not supported",
+		"not supported",
+		"same regs are invalid",
+		"internal error"
+	};
+	assert(ERR_INTERNAL + 1 == sizeof(errTbl) / sizeof(*errTbl));
+	return err <= ERR_INTERNAL ? errTbl[err] : "unknown err";
+}
+
+#ifdef XBYAK_NO_EXCEPTION
+namespace local {
+
+inline int& GetErrorRef() {
+	static XBYAK_TLS int err = 0;
+	return err;
+}
+
+inline void SetError(int err) {
+	if (local::GetErrorRef()) return; // keep the first err code
+	local::GetErrorRef() = err;
+}
+
+} // local
+
+inline void ClearError() {
+	local::GetErrorRef() = 0;
+}
+inline int GetError() { return Xbyak::local::GetErrorRef(); }
+
+#define XBYAK_THROW(err) { Xbyak::local::SetError(err); return; }
+#define XBYAK_THROW_RET(err, r) { Xbyak::local::SetError(err); return r; }
+
+#else
 class Error : public std::exception {
 	int err_;
 public:
 	explicit Error(int err) : err_(err)
 	{
 		if (err_ < 0 || err_ > ERR_INTERNAL) {
-			fprintf(stderr, "bad err=%d in Xbyak::Error\n", err_);
-			exit(1);
+			err_ = ERR_INTERNAL;
 		}
 	}
 	operator int() const { return err_; }
-	const char *what() const throw()
-	{
-		static const char *errTbl[] = {
-			"none",
-			"bad addressing",
-			"code is too big",
-			"bad scale",
-			"esp can't be index",
-			"bad combination",
-			"bad size of register",
-			"imm is too big",
-			"bad align",
-			"label is redefined",
-			"label is too far",
-			"label is not found",
-			"code is not copyable",
-			"bad parameter",
-			"can't protect",
-			"can't use 64bit disp(use (void*))",
-			"offset is too big",
-			"MEM size is not specified",
-			"bad mem size",
-			"bad st combination",
-			"over local label",
-			"under local label",
-			"can't alloc",
-			"T_SHORT is not supported in AutoGrow",
-			"bad protect mode",
-			"bad pNum",
-			"bad tNum",
-			"bad vsib addressing",
-			"can't convert",
-			"label is not set by L()",
-			"label is already set by L()",
-			"bad label string",
-			"err munmap",
-			"opmask is already set",
-			"rounding is already set",
-			"k0 is invalid",
-			"evex is invalid",
-			"sae(suppress all exceptions) is invalid",
-			"er(embedded rounding) is invalid",
-			"invalid broadcast",
-			"invalid opmask with memory",
-			"invalid zero",
-			"invalid rip in AutoGrow",
-			"internal error",
-		};
-		assert((size_t)err_ < sizeof(errTbl) / sizeof(*errTbl));
-		return errTbl[err_];
+	const char *what() const XBYAK_NOEXCEPT
+	{
+		return ConvertErrorToString(err_);
 	}
 };
 
-inline const char *ConvertErrorToString(Error err)
+// dummy functions
+inline void ClearError() { }
+inline int GetError() { return 0; }
+
+inline const char *ConvertErrorToString(const Error& err)
 {
 	return err.what();
 }
 
+#define XBYAK_THROW(err) { throw Error(err); }
+#define XBYAK_THROW_RET(err, r) { throw Error(err); }
+
+#endif
+
 inline void *AlignedMalloc(size_t size, size_t alignment)
 {
 #ifdef __MINGW32__
@@ -274,23 +365,45 @@ inline void AlignedFree(void *p)
 }
 
 template<class To, class From>
-inline const To CastTo(From p) throw()
+inline const To CastTo(From p) XBYAK_NOEXCEPT
 {
 	return (const To)(size_t)(p);
 }
 namespace inner {
 
-static const size_t ALIGN_PAGE_SIZE = 4096;
+#ifdef _WIN32
+struct SystemInfo {
+	SYSTEM_INFO info;
+	SystemInfo()
+	{
+		GetSystemInfo(&info);
+	}
+};
+#endif
+//static const size_t ALIGN_PAGE_SIZE = 4096;
+inline size_t getPageSize()
+{
+#ifdef _WIN32
+	static const SystemInfo si;
+	return si.info.dwPageSize;
+#elif defined(__GNUC__)
+	static const long pageSize = sysconf(_SC_PAGESIZE);
+	if (pageSize > 0) {
+		return (size_t)pageSize;
+	}
+#endif
+	return 4096;
+}
 
-inline bool IsInDisp8(uint32 x) { return 0xFFFFFF80 <= x || x <= 0x7F; }
-inline bool IsInInt32(uint64 x) { return ~uint64(0x7fffffffu) <= x || x <= 0x7FFFFFFFU; }
+inline bool IsInDisp8(uint32_t x) { return 0xFFFFFF80 <= x || x <= 0x7F; }
+inline bool IsInInt32(uint64_t x) { return ~uint64_t(0x7fffffffu) <= x || x <= 0x7FFFFFFFU; }
 
-inline uint32 VerifyInInt32(uint64 x)
+inline uint32_t VerifyInInt32(uint64_t x)
 {
-#ifdef XBYAK64
-	if (!IsInInt32(x)) throw Error(ERR_OFFSET_IS_TOO_BIG);
+#if defined(XBYAK64) && !defined(__ILP32__)
+	if (!IsInInt32(x)) XBYAK_THROW_RET(ERR_OFFSET_IS_TOO_BIG, 0)
 #endif
-	return static_cast<uint32>(x);
+	return static_cast<uint32_t>(x);
 }
 
 enum LabelMode {
@@ -305,51 +418,116 @@ enum LabelMode {
 	custom allocator
 */
 struct Allocator {
-	virtual uint8 *alloc(size_t size) { return reinterpret_cast<uint8*>(AlignedMalloc(size, inner::ALIGN_PAGE_SIZE)); }
-	virtual void free(uint8 *p) { AlignedFree(p); }
+	explicit Allocator(const std::string& = "") {} // same interface with MmapAllocator
+	virtual uint8_t *alloc(size_t size) { return reinterpret_cast<uint8_t*>(AlignedMalloc(size, inner::getPageSize())); }
+	virtual void free(uint8_t *p) { AlignedFree(p); }
 	virtual ~Allocator() {}
 	/* override to return false if you call protect() manually */
 	virtual bool useProtect() const { return true; }
 };
 
 #ifdef XBYAK_USE_MMAP_ALLOCATOR
-class MmapAllocator : Allocator {
-	typedef XBYAK_STD_UNORDERED_MAP<uintptr_t, size_t> SizeList;
-	SizeList sizeList_;
+#ifdef XBYAK_USE_MAP_JIT
+namespace util {
+
+inline int getMacOsVersionPure()
+{
+	char buf[64];
+	size_t size = sizeof(buf);
+	int err = sysctlbyname("kern.osrelease", buf, &size, NULL, 0);
+	if (err != 0) return 0;
+	char *endp;
+	int major = strtol(buf, &endp, 10);
+	if (*endp != '.') return 0;
+	return major;
+}
+
+inline int getMacOsVersion()
+{
+	static const int version = getMacOsVersionPure();
+	return version;
+}
+
+} // util
+#endif
+class MmapAllocator : public Allocator {
+	struct Allocation {
+		size_t size;
+#if defined(XBYAK_USE_MEMFD)
+		// fd_ is only used with XBYAK_USE_MEMFD. We keep the file open
+		// during the lifetime of each allocation in order to support
+		// checkpoint/restore by unprivileged users.
+		int fd;
+#endif
+	};
+	const std::string name_; // only used with XBYAK_USE_MEMFD
+	typedef XBYAK_STD_UNORDERED_MAP<uintptr_t, Allocation> AllocationList;
+	AllocationList allocList_;
 public:
-	uint8 *alloc(size_t size)
+	explicit MmapAllocator(const std::string& name = "xbyak") : name_(name) {}
+	uint8_t *alloc(size_t size)
 	{
-		const size_t alignedSizeM1 = inner::ALIGN_PAGE_SIZE - 1;
+		const size_t alignedSizeM1 = inner::getPageSize() - 1;
 		size = (size + alignedSizeM1) & ~alignedSizeM1;
-#ifdef MAP_ANONYMOUS
-		const int mode = MAP_PRIVATE | MAP_ANONYMOUS;
+#if defined(MAP_ANONYMOUS)
+		int mode = MAP_PRIVATE | MAP_ANONYMOUS;
 #elif defined(MAP_ANON)
-		const int mode = MAP_PRIVATE | MAP_ANON;
+		int mode = MAP_PRIVATE | MAP_ANON;
 #else
 		#error "not supported"
 #endif
-		void *p = mmap(NULL, size, PROT_READ | PROT_WRITE, mode, -1, 0);
-		if (p == MAP_FAILED) throw Error(ERR_CANT_ALLOC);
+#if defined(XBYAK_USE_MAP_JIT)
+		const int mojaveVersion = 18;
+		if (util::getMacOsVersion() >= mojaveVersion) mode |= MAP_JIT;
+#endif
+		int fd = -1;
+#if defined(XBYAK_USE_MEMFD)
+		fd = memfd_create(name_.c_str(), MFD_CLOEXEC);
+		if (fd != -1) {
+			mode = MAP_SHARED;
+			if (ftruncate(fd, size) != 0) {
+				close(fd);
+				XBYAK_THROW_RET(ERR_CANT_ALLOC, 0)
+			}
+		}
+#endif
+		void *p = mmap(NULL, size, PROT_READ | PROT_WRITE, mode, fd, 0);
+		if (p == MAP_FAILED) {
+			if (fd != -1) close(fd);
+			XBYAK_THROW_RET(ERR_CANT_ALLOC, 0)
+		}
 		assert(p);
-		sizeList_[(uintptr_t)p] = size;
-		return (uint8*)p;
+		Allocation &alloc = allocList_[(uintptr_t)p];
+		alloc.size = size;
+#if defined(XBYAK_USE_MEMFD)
+		alloc.fd = fd;
+#endif
+		return (uint8_t*)p;
 	}
-	void free(uint8 *p)
+	void free(uint8_t *p)
 	{
 		if (p == 0) return;
-		SizeList::iterator i = sizeList_.find((uintptr_t)p);
-		if (i == sizeList_.end()) throw Error(ERR_BAD_PARAMETER);
-		if (munmap((void*)i->first, i->second) < 0) throw Error(ERR_MUNMAP);
-		sizeList_.erase(i);
+		AllocationList::iterator i = allocList_.find((uintptr_t)p);
+		if (i == allocList_.end()) XBYAK_THROW(ERR_BAD_PARAMETER)
+		if (munmap((void*)i->first, i->second.size) < 0) XBYAK_THROW(ERR_MUNMAP)
+#if defined(XBYAK_USE_MEMFD)
+		if (i->second.fd != -1) close(i->second.fd);
+#endif
+		allocList_.erase(i);
 	}
 };
+#else
+typedef Allocator MmapAllocator;
 #endif
 
+class Address;
+class Reg;
+
 class Operand {
-	static const uint8 EXT8BIT = 0x40;
-	unsigned int idx_:7; // 0..31, EXT8BIT = 1 if spl/bpl/sil/dil
-	unsigned int kind_:8;
-	unsigned int bit_:10;
+	static const uint8_t EXT8BIT = 0x20;
+	unsigned int idx_:6; // 0..31 + EXT8BIT = 1 if spl/bpl/sil/dil
+	unsigned int kind_:10;
+	unsigned int bit_:14;
 protected:
 	unsigned int zero_:1;
 	unsigned int mask_:3;
@@ -365,7 +543,9 @@ public:
 		XMM = 1 << 4,
 		YMM = 1 << 5,
 		ZMM = 1 << 6,
-		OPMASK = 1 << 7
+		OPMASK = 1 << 7,
+		BNDREG = 1 << 8,
+		TMM = 1 << 9
 	};
 	enum Code {
 #ifdef XBYAK64
@@ -379,53 +559,55 @@ public:
 		AX = 0, CX, DX, BX, SP, BP, SI, DI,
 		AL = 0, CL, DL, BL, AH, CH, DH, BH
 	};
-	Operand() : idx_(0), kind_(0), bit_(0), zero_(0), mask_(0), rounding_(0) { }
-	Operand(int idx, Kind kind, int bit, bool ext8bit = 0)
-		: idx_(static_cast<uint8>(idx | (ext8bit ? EXT8BIT : 0)))
-		, kind_(static_cast<uint8>(kind))
+	XBYAK_CONSTEXPR Operand() : idx_(0), kind_(0), bit_(0), zero_(0), mask_(0), rounding_(0) { }
+	XBYAK_CONSTEXPR Operand(int idx, Kind kind, int bit, bool ext8bit = 0)
+		: idx_(static_cast<uint8_t>(idx | (ext8bit ? EXT8BIT : 0)))
+		, kind_(kind)
 		, bit_(bit)
 		, zero_(0), mask_(0), rounding_(0)
 	{
 		assert((bit_ & (bit_ - 1)) == 0); // bit must be power of two
 	}
-	Kind getKind() const { return static_cast<Kind>(kind_); }
-	int getIdx() const { return idx_ & (EXT8BIT - 1); }
-	bool isNone() const { return kind_ == 0; }
-	bool isMMX() const { return is(MMX); }
-	bool isXMM() const { return is(XMM); }
-	bool isYMM() const { return is(YMM); }
-	bool isZMM() const { return is(ZMM); }
-	bool isXMEM() const { return is(XMM | MEM); }
-	bool isYMEM() const { return is(YMM | MEM); }
-	bool isZMEM() const { return is(ZMM | MEM); }
-	bool isOPMASK() const { return is(OPMASK); }
-	bool isREG(int bit = 0) const { return is(REG, bit); }
-	bool isMEM(int bit = 0) const { return is(MEM, bit); }
-	bool isFPU() const { return is(FPU); }
-	bool isExt8bit() const { return (idx_ & EXT8BIT) != 0; }
-	bool isExtIdx() const { return (getIdx() & 8) != 0; }
-	bool isExtIdx2() const { return (getIdx() & 16) != 0; }
-	bool hasEvex() const { return isZMM() || isExtIdx2() || hasZero() || getOpmaskIdx() || getRounding(); }
-	bool hasRex() const { return isExt8bit() | isREG(64) | isExtIdx(); }
-	bool hasZero() const { return zero_; }
-	int getOpmaskIdx() const { return mask_; }
-	int getRounding() const { return rounding_; }
+	XBYAK_CONSTEXPR Kind getKind() const { return static_cast<Kind>(kind_); }
+	XBYAK_CONSTEXPR int getIdx() const { return idx_ & (EXT8BIT - 1); }
+	XBYAK_CONSTEXPR bool isNone() const { return kind_ == 0; }
+	XBYAK_CONSTEXPR bool isMMX() const { return is(MMX); }
+	XBYAK_CONSTEXPR bool isXMM() const { return is(XMM); }
+	XBYAK_CONSTEXPR bool isYMM() const { return is(YMM); }
+	XBYAK_CONSTEXPR bool isZMM() const { return is(ZMM); }
+	XBYAK_CONSTEXPR bool isTMM() const { return is(TMM); }
+	XBYAK_CONSTEXPR bool isXMEM() const { return is(XMM | MEM); }
+	XBYAK_CONSTEXPR bool isYMEM() const { return is(YMM | MEM); }
+	XBYAK_CONSTEXPR bool isZMEM() const { return is(ZMM | MEM); }
+	XBYAK_CONSTEXPR bool isOPMASK() const { return is(OPMASK); }
+	XBYAK_CONSTEXPR bool isBNDREG() const { return is(BNDREG); }
+	XBYAK_CONSTEXPR bool isREG(int bit = 0) const { return is(REG, bit); }
+	XBYAK_CONSTEXPR bool isMEM(int bit = 0) const { return is(MEM, bit); }
+	XBYAK_CONSTEXPR bool isFPU() const { return is(FPU); }
+	XBYAK_CONSTEXPR bool isExt8bit() const { return (idx_ & EXT8BIT) != 0; }
+	XBYAK_CONSTEXPR bool isExtIdx() const { return (getIdx() & 8) != 0; }
+	XBYAK_CONSTEXPR bool isExtIdx2() const { return (getIdx() & 16) != 0; }
+	XBYAK_CONSTEXPR bool hasEvex() const { return isZMM() || isExtIdx2() || getOpmaskIdx() || getRounding(); }
+	XBYAK_CONSTEXPR bool hasRex() const { return isExt8bit() || isREG(64) || isExtIdx(); }
+	XBYAK_CONSTEXPR bool hasZero() const { return zero_; }
+	XBYAK_CONSTEXPR int getOpmaskIdx() const { return mask_; }
+	XBYAK_CONSTEXPR int getRounding() const { return rounding_; }
 	void setKind(Kind kind)
 	{
-		if ((kind & (XMM|YMM|ZMM)) == 0) return;
+		if ((kind & (XMM|YMM|ZMM|TMM)) == 0) return;
 		kind_ = kind;
-		bit_ = kind == XMM ? 128 : kind == YMM ? 256 : 512;
+		bit_ = kind == XMM ? 128 : kind == YMM ? 256 : kind == ZMM ? 512 : 8192;
 	}
-	void setBit(int bit) { bit_ = bit; }
-	void setOpmaskIdx(int idx, bool ignore_idx0 = false)
+	// err if MMX/FPU/OPMASK/BNDREG
+	void setBit(int bit);
+	void setOpmaskIdx(int idx, bool /*ignore_idx0*/ = true)
 	{
-		if (!ignore_idx0 && idx == 0) throw Error(ERR_K0_IS_INVALID);
-		if (mask_) throw Error(ERR_OPMASK_IS_ALREADY_SET);
+		if (mask_) XBYAK_THROW(ERR_OPMASK_IS_ALREADY_SET)
 		mask_ = idx;
 	}
 	void setRounding(int idx)
 	{
-		if (rounding_) throw Error(ERR_ROUNDING_IS_ALREADY_SET);
+		if (rounding_) XBYAK_THROW(ERR_ROUNDING_IS_ALREADY_SET)
 		rounding_ = idx;
 	}
 	void setZero() { zero_ = true; }
@@ -438,12 +620,12 @@ public:
 		return AH <= idx && idx <= BH;
 	}
 	// any bit is accetable if bit == 0
-	bool is(int kind, uint32 bit = 0) const
+	XBYAK_CONSTEXPR bool is(int kind, uint32_t bit = 0) const
 	{
 		return (kind == 0 || (kind_ & kind)) && (bit == 0 || (bit_ & bit)); // cf. you can set (8|16)
 	}
-	bool isBit(uint32 bit) const { return (bit_ & bit) != 0; }
-	uint32 getBit() const { return bit_; }
+	XBYAK_CONSTEXPR bool isBit(uint32_t bit) const { return (bit_ & bit) != 0; }
+	XBYAK_CONSTEXPR uint32_t getBit() const { return bit_; }
 	const char *toString() const
 	{
 		const int idx = getIdx();
@@ -462,6 +644,11 @@ public:
 		} else if (isOPMASK()) {
 			static const char *tbl[8] = { "k0", "k1", "k2", "k3", "k4", "k5", "k6", "k7" };
 			return tbl[idx];
+		} else if (isTMM()) {
+			static const char *tbl[8] = {
+				"tmm0", "tmm1", "tmm2", "tmm3", "tmm4", "tmm5", "tmm6", "tmm7"
+			};
+			return tbl[idx];
 		} else if (isZMM()) {
 			static const char *tbl[32] = {
 				"zmm0", "zmm1", "zmm2", "zmm3", "zmm4", "zmm5", "zmm6", "zmm7", "zmm8", "zmm9", "zmm10", "zmm11", "zmm12", "zmm13", "zmm14", "zmm15",
@@ -486,14 +673,63 @@ public:
 		} else if (isFPU()) {
 			static const char *tbl[8] = { "st0", "st1", "st2", "st3", "st4", "st5", "st6", "st7" };
 			return tbl[idx];
+		} else if (isBNDREG()) {
+			static const char *tbl[4] = { "bnd0", "bnd1", "bnd2", "bnd3" };
+			return tbl[idx];
 		}
-		throw Error(ERR_INTERNAL);
+		XBYAK_THROW_RET(ERR_INTERNAL, 0);
 	}
 	bool isEqualIfNotInherited(const Operand& rhs) const { return idx_ == rhs.idx_ && kind_ == rhs.kind_ && bit_ == rhs.bit_ && zero_ == rhs.zero_ && mask_ == rhs.mask_ && rounding_ == rhs.rounding_; }
 	bool operator==(const Operand& rhs) const;
 	bool operator!=(const Operand& rhs) const { return !operator==(rhs); }
+	const Address& getAddress() const;
+	const Reg& getReg() const;
 };
 
+inline void Operand::setBit(int bit)
+{
+	if (bit != 8 && bit != 16 && bit != 32 && bit != 64 && bit != 128 && bit != 256 && bit != 512 && bit != 8192) goto ERR;
+	if (isBit(bit)) return;
+	if (is(MEM | OPMASK)) {
+		bit_ = bit;
+		return;
+	}
+	if (is(REG | XMM | YMM | ZMM | TMM)) {
+		int idx = getIdx();
+		// err if converting ah, bh, ch, dh
+		if (isREG(8) && (4 <= idx && idx < 8) && !isExt8bit()) goto ERR;
+		Kind kind = REG;
+		switch (bit) {
+		case 8:
+			if (idx >= 16) goto ERR;
+#ifdef XBYAK32
+			if (idx >= 4) goto ERR;
+#else
+			if (4 <= idx && idx < 8) idx |= EXT8BIT;
+#endif
+			break;
+		case 16:
+		case 32:
+		case 64:
+			if (idx >= 16) goto ERR;
+			break;
+		case 128: kind = XMM; break;
+		case 256: kind = YMM; break;
+		case 512: kind = ZMM; break;
+		case 8192: kind = TMM; break;
+		}
+		idx_ = idx;
+		kind_ = kind;
+		bit_ = bit;
+		if (bit >= 128) return; // keep mask_ and rounding_
+		mask_ = 0;
+		rounding_ = 0;
+		return;
+	}
+ERR:
+	XBYAK_THROW(ERR_CANT_CONVERT)
+}
+
 class Label;
 
 struct Reg8;
@@ -504,16 +740,17 @@ struct Reg64;
 #endif
 class Reg : public Operand {
 public:
-	Reg() { }
-	Reg(int idx, Kind kind, int bit = 0, bool ext8bit = false) : Operand(idx, kind, bit, ext8bit) { }
-	Reg changeBit(int bit) const { return Reg(getIdx(), getKind(), bit, isExt8bit()); }
-	uint8 getRexW() const { return isREG(64) ? 8 : 0; }
-	uint8 getRexR() const { return isExtIdx() ? 4 : 0; }
-	uint8 getRexX() const { return isExtIdx() ? 2 : 0; }
-	uint8 getRexB() const { return isExtIdx() ? 1 : 0; }
-	uint8 getRex(const Reg& base = Reg()) const
-	{
-		uint8 rex = getRexW() | getRexR() | base.getRexW() | base.getRexB();
+	XBYAK_CONSTEXPR Reg() { }
+	XBYAK_CONSTEXPR Reg(int idx, Kind kind, int bit = 0, bool ext8bit = false) : Operand(idx, kind, bit, ext8bit) { }
+	// convert to Reg8/Reg16/Reg32/Reg64/XMM/YMM/ZMM
+	Reg changeBit(int bit) const { Reg r(*this); r.setBit(bit); return r; }
+	uint8_t getRexW() const { return isREG(64) ? 8 : 0; }
+	uint8_t getRexR() const { return isExtIdx() ? 4 : 0; }
+	uint8_t getRexX() const { return isExtIdx() ? 2 : 0; }
+	uint8_t getRexB() const { return isExtIdx() ? 1 : 0; }
+	uint8_t getRex(const Reg& base = Reg()) const
+	{
+		uint8_t rex = getRexW() | getRexR() | base.getRexW() | base.getRexB();
 		if (rex || isExt8bit() || base.isExt8bit()) rex |= 0x40;
 		return rex;
 	}
@@ -525,44 +762,67 @@ public:
 #endif
 };
 
+inline const Reg& Operand::getReg() const
+{
+	assert(!isMEM());
+	return static_cast<const Reg&>(*this);
+}
+
 struct Reg8 : public Reg {
-	explicit Reg8(int idx = 0, bool ext8bit = false) : Reg(idx, Operand::REG, 8, ext8bit) { }
+	explicit XBYAK_CONSTEXPR Reg8(int idx = 0, bool ext8bit = false) : Reg(idx, Operand::REG, 8, ext8bit) { }
 };
 
 struct Reg16 : public Reg {
-	explicit Reg16(int idx = 0) : Reg(idx, Operand::REG, 16) { }
+	explicit XBYAK_CONSTEXPR Reg16(int idx = 0) : Reg(idx, Operand::REG, 16) { }
 };
 
 struct Mmx : public Reg {
-	explicit Mmx(int idx = 0, Kind kind = Operand::MMX, int bit = 64) : Reg(idx, kind, bit) { }
+	explicit XBYAK_CONSTEXPR Mmx(int idx = 0, Kind kind = Operand::MMX, int bit = 64) : Reg(idx, kind, bit) { }
 };
 
 struct EvexModifierRounding {
-	explicit EvexModifierRounding(int rounding) : rounding(rounding) {}
+	enum {
+		T_RN_SAE = 1,
+		T_RD_SAE = 2,
+		T_RU_SAE = 3,
+		T_RZ_SAE = 4,
+		T_SAE = 5
+	};
+	explicit XBYAK_CONSTEXPR EvexModifierRounding(int rounding) : rounding(rounding) {}
 	int rounding;
 };
-struct EvexModifierZero{};
+struct EvexModifierZero{ XBYAK_CONSTEXPR EvexModifierZero() {}};
 
 struct Xmm : public Mmx {
-	explicit Xmm(int idx = 0, Kind kind = Operand::XMM, int bit = 128) : Mmx(idx, kind, bit) { }
-	Xmm(Kind kind, int idx) : Mmx(idx, kind, kind == XMM ? 128 : kind == YMM ? 256 : 512) { }
+	explicit XBYAK_CONSTEXPR Xmm(int idx = 0, Kind kind = Operand::XMM, int bit = 128) : Mmx(idx, kind, bit) { }
+	XBYAK_CONSTEXPR Xmm(Kind kind, int idx) : Mmx(idx, kind, kind == XMM ? 128 : kind == YMM ? 256 : 512) { }
 	Xmm operator|(const EvexModifierRounding& emr) const { Xmm r(*this); r.setRounding(emr.rounding); return r; }
 	Xmm copyAndSetIdx(int idx) const { Xmm ret(*this); ret.setIdx(idx); return ret; }
 	Xmm copyAndSetKind(Operand::Kind kind) const { Xmm ret(*this); ret.setKind(kind); return ret; }
 };
 
 struct Ymm : public Xmm {
-	explicit Ymm(int idx = 0, Kind kind = Operand::YMM, int bit = 256) : Xmm(idx, kind, bit) { }
+	explicit XBYAK_CONSTEXPR Ymm(int idx = 0, Kind kind = Operand::YMM, int bit = 256) : Xmm(idx, kind, bit) { }
 	Ymm operator|(const EvexModifierRounding& emr) const { Ymm r(*this); r.setRounding(emr.rounding); return r; }
 };
 
 struct Zmm : public Ymm {
-	explicit Zmm(int idx = 0) : Ymm(idx, Operand::ZMM, 512) { }
+	explicit XBYAK_CONSTEXPR Zmm(int idx = 0) : Ymm(idx, Operand::ZMM, 512) { }
 	Zmm operator|(const EvexModifierRounding& emr) const { Zmm r(*this); r.setRounding(emr.rounding); return r; }
 };
 
+#ifdef XBYAK64
+struct Tmm : public Reg {
+	explicit XBYAK_CONSTEXPR Tmm(int idx = 0, Kind kind = Operand::TMM, int bit = 8192) : Reg(idx, kind, bit) { }
+};
+#endif
+
 struct Opmask : public Reg {
-	explicit Opmask(int idx = 0) : Reg(idx, Operand::OPMASK, 64) {}
+	explicit XBYAK_CONSTEXPR Opmask(int idx = 0) : Reg(idx, Operand::OPMASK, 64) {}
+};
+
+struct BoundsReg : public Reg {
+	explicit XBYAK_CONSTEXPR BoundsReg(int idx = 0) : Reg(idx, Operand::BNDREG, 128) {}
 };
 
 template<class T>T operator|(const T& x, const Opmask& k) { T r(x); r.setOpmaskIdx(k.getIdx()); return r; }
@@ -570,71 +830,66 @@ template<class T>T operator|(const T& x, const EvexModifierZero&) { T r(x); r.se
 template<class T>T operator|(const T& x, const EvexModifierRounding& emr) { T r(x); r.setRounding(emr.rounding); return r; }
 
 struct Fpu : public Reg {
-	explicit Fpu(int idx = 0) : Reg(idx, Operand::FPU, 32) { }
+	explicit XBYAK_CONSTEXPR Fpu(int idx = 0) : Reg(idx, Operand::FPU, 32) { }
 };
 
 struct Reg32e : public Reg {
-	explicit Reg32e(int idx, int bit) : Reg(idx, Operand::REG, bit) {}
+	explicit XBYAK_CONSTEXPR Reg32e(int idx, int bit) : Reg(idx, Operand::REG, bit) {}
 };
 struct Reg32 : public Reg32e {
-	explicit Reg32(int idx = 0) : Reg32e(idx, 32) {}
+	explicit XBYAK_CONSTEXPR Reg32(int idx = 0) : Reg32e(idx, 32) {}
 };
 #ifdef XBYAK64
 struct Reg64 : public Reg32e {
-	explicit Reg64(int idx = 0) : Reg32e(idx, 64) {}
+	explicit XBYAK_CONSTEXPR Reg64(int idx = 0) : Reg32e(idx, 64) {}
 };
 struct RegRip {
-	sint64 disp_;
-	Label* label_;
+	int64_t disp_;
+	const Label* label_;
 	bool isAddr_;
-	explicit RegRip(sint64 disp = 0, Label* label = 0, bool isAddr = false) : disp_(disp), label_(label), isAddr_(isAddr) {}
-	friend const RegRip operator+(const RegRip& r, sint64 disp) {
+	explicit XBYAK_CONSTEXPR RegRip(int64_t disp = 0, const Label* label = 0, bool isAddr = false) : disp_(disp), label_(label), isAddr_(isAddr) {}
+	friend const RegRip operator+(const RegRip& r, int disp) {
+		return RegRip(r.disp_ + disp, r.label_, r.isAddr_);
+	}
+	friend const RegRip operator-(const RegRip& r, int disp) {
+		return RegRip(r.disp_ - disp, r.label_, r.isAddr_);
+	}
+	friend const RegRip operator+(const RegRip& r, int64_t disp) {
 		return RegRip(r.disp_ + disp, r.label_, r.isAddr_);
 	}
-	friend const RegRip operator-(const RegRip& r, sint64 disp) {
+	friend const RegRip operator-(const RegRip& r, int64_t disp) {
 		return RegRip(r.disp_ - disp, r.label_, r.isAddr_);
 	}
-	friend const RegRip operator+(const RegRip& r, Label& label) {
-		if (r.label_ || r.isAddr_) throw Error(ERR_BAD_ADDRESSING);
+	friend const RegRip operator+(const RegRip& r, const Label& label) {
+		if (r.label_ || r.isAddr_) XBYAK_THROW_RET(ERR_BAD_ADDRESSING, RegRip());
 		return RegRip(r.disp_, &label);
 	}
 	friend const RegRip operator+(const RegRip& r, const void *addr) {
-		if (r.label_ || r.isAddr_) throw Error(ERR_BAD_ADDRESSING);
-		return RegRip(r.disp_ + (sint64)addr, 0, true);
+		if (r.label_ || r.isAddr_) XBYAK_THROW_RET(ERR_BAD_ADDRESSING, RegRip());
+		return RegRip(r.disp_ + (int64_t)addr, 0, true);
 	}
 };
 #endif
 
 inline Reg8 Reg::cvt8() const
 {
-	const int idx = getIdx();
-	if (isBit(8)) return Reg8(idx, isExt8bit());
-#ifdef XBYAK32
-	if (idx >= 4) throw Error(ERR_CANT_CONVERT);
-#endif
-	return Reg8(idx, 4 <= idx && idx < 8);
+	Reg r = changeBit(8); return Reg8(r.getIdx(), r.isExt8bit());
 }
 
 inline Reg16 Reg::cvt16() const
 {
-	const int idx = getIdx();
-	if (isBit(8) && (4 <= idx && idx < 8) && !isExt8bit()) throw Error(ERR_CANT_CONVERT);
-	return Reg16(idx);
+	return Reg16(changeBit(16).getIdx());
 }
 
 inline Reg32 Reg::cvt32() const
 {
-	const int idx = getIdx();
-	if (isBit(8) && (4 <= idx && idx < 8) && !isExt8bit()) throw Error(ERR_CANT_CONVERT);
-	return Reg32(idx);
+	return Reg32(changeBit(32).getIdx());
 }
 
 #ifdef XBYAK64
 inline Reg64 Reg::cvt64() const
 {
-	const int idx = getIdx();
-	if (isBit(8) && (4 <= idx && idx < 8) && !isExt8bit()) throw Error(ERR_CANT_CONVERT);
-	return Reg64(idx);
+	return Reg64(changeBit(64).getIdx());
 }
 #endif
 
@@ -646,7 +901,7 @@ public:
 	enum {
 		es, cs, ss, ds, fs, gs
 	};
-	explicit Segment(int idx) : idx_(idx) { assert(0 <= idx_ && idx_ < 6); }
+	explicit XBYAK_CONSTEXPR Segment(int idx) : idx_(idx) { assert(0 <= idx_ && idx_ < 6); }
 	int getIdx() const { return idx_; }
 	const char *toString() const
 	{
@@ -665,14 +920,14 @@ public:
 #else
 	enum { i32e = 32 };
 #endif
-	RegExp(size_t disp = 0) : scale_(0), disp_(disp) { }
-	RegExp(const Reg& r, int scale = 1)
+	XBYAK_CONSTEXPR RegExp(size_t disp = 0) : scale_(0), disp_(disp) { }
+	XBYAK_CONSTEXPR RegExp(const Reg& r, int scale = 1)
 		: scale_(scale)
 		, disp_(0)
 	{
-		if (!r.isREG(i32e) && !r.is(Reg::XMM|Reg::YMM|Reg::ZMM)) throw Error(ERR_BAD_SIZE_OF_REGISTER);
+		if (!r.isREG(i32e) && !r.is(Reg::XMM|Reg::YMM|Reg::ZMM|Reg::TMM)) XBYAK_THROW(ERR_BAD_SIZE_OF_REGISTER)
 		if (scale == 0) return;
-		if (scale != 1 && scale != 2 && scale != 4 && scale != 8) throw Error(ERR_BAD_SCALE);
+		if (scale != 1 && scale != 2 && scale != 4 && scale != 8) XBYAK_THROW(ERR_BAD_SCALE)
 		if (r.getBit() >= 128 || scale != 1) { // xmm/ymm is always index
 			index_ = r;
 		} else {
@@ -680,13 +935,15 @@ public:
 		}
 	}
 	bool isVsib(int bit = 128 | 256 | 512) const { return index_.isBit(bit); }
-	void optimize()
+	RegExp optimize() const
 	{
+		RegExp exp = *this;
 		// [reg * 2] => [reg + reg]
-		if (index_.isBit(i32e) && !base_.getBit() && index_.getBit() && scale_ == 2) {
-			base_ = index_;
-			scale_ = 1;
+		if (index_.isBit(i32e) && !base_.getBit() && scale_ == 2) {
+			exp.base_ = index_;
+			exp.scale_ = 1;
 		}
+		return exp;
 	}
 	bool operator==(const RegExp& rhs) const
 	{
@@ -696,16 +953,21 @@ public:
 	const Reg& getIndex() const { return index_; }
 	int getScale() const { return scale_; }
 	size_t getDisp() const { return disp_; }
-	void verify() const
+	XBYAK_CONSTEXPR void verify() const
 	{
-		if (base_.getBit() >= 128) throw Error(ERR_BAD_SIZE_OF_REGISTER);
+		if (base_.getBit() >= 128) XBYAK_THROW(ERR_BAD_SIZE_OF_REGISTER)
 		if (index_.getBit() && index_.getBit() <= 64) {
-			if (index_.getIdx() == Operand::ESP) throw Error(ERR_ESP_CANT_BE_INDEX);
-			if (base_.getBit() && base_.getBit() != index_.getBit()) throw Error(ERR_BAD_SIZE_OF_REGISTER);
+			if (index_.getIdx() == Operand::ESP) XBYAK_THROW(ERR_ESP_CANT_BE_INDEX)
+			if (base_.getBit() && base_.getBit() != index_.getBit()) XBYAK_THROW(ERR_BAD_SIZE_OF_REGISTER)
 		}
 	}
 	friend RegExp operator+(const RegExp& a, const RegExp& b);
 	friend RegExp operator-(const RegExp& e, size_t disp);
+	uint8_t getRex() const
+	{
+		uint8_t rex = index_.getRexX() | base_.getRexB();
+		return rex ? uint8_t(rex | 0x40) : 0;
+	}
 private:
 	/*
 		[base_ + index_ * scale_ + disp_]
@@ -719,12 +981,12 @@ private:
 
 inline RegExp operator+(const RegExp& a, const RegExp& b)
 {
-	if (a.index_.getBit() && b.index_.getBit()) throw Error(ERR_BAD_ADDRESSING);
+	if (a.index_.getBit() && b.index_.getBit()) XBYAK_THROW_RET(ERR_BAD_ADDRESSING, RegExp())
 	RegExp ret = a;
 	if (!ret.index_.getBit()) { ret.index_ = b.index_; ret.scale_ = b.scale_; }
 	if (b.base_.getBit()) {
 		if (ret.base_.getBit()) {
-			if (ret.index_.getBit()) throw Error(ERR_BAD_ADDRESSING);
+			if (ret.index_.getBit()) XBYAK_THROW_RET(ERR_BAD_ADDRESSING, RegExp())
 			// base + base => base + index * 1
 			ret.index_ = b.base_;
 			// [reg + esp] => [esp + reg]
@@ -741,6 +1003,10 @@ inline RegExp operator*(const Reg& r, int scale)
 {
 	return RegExp(r, scale);
 }
+inline RegExp operator*(int scale, const Reg& r)
+{
+	return r * scale;
+}
 inline RegExp operator-(const RegExp& e, size_t disp)
 {
 	RegExp ret = e;
@@ -749,7 +1015,8 @@ inline RegExp operator-(const RegExp& e, size_t disp)
 }
 
 // 2nd parameter for constructor of CodeArray(maxSize, userPtr, alloc)
-void *const AutoGrow = (void*)1;
+void *const AutoGrow = (void*)1; //-V566
+void *const DontSetProtectRWE = (void*)2; //-V566
 
 class CodeArray {
 	enum Type {
@@ -767,9 +1034,9 @@ class CodeArray {
 		inner::LabelMode mode;
 		AddrInfo(size_t _codeOffset, size_t _jmpAddr, int _jmpSize, inner::LabelMode _mode)
 			: codeOffset(_codeOffset), jmpAddr(_jmpAddr), jmpSize(_jmpSize), mode(_mode) {}
-		uint64 getVal(const uint8 *top) const
+		uint64_t getVal(const uint8_t *top) const
 		{
-			uint64 disp = (mode == inner::LaddTop) ? jmpAddr + size_t(top) : (mode == inner::LasIs) ? jmpAddr : jmpAddr - size_t(top);
+			uint64_t disp = (mode == inner::LaddTop) ? jmpAddr + size_t(top) : (mode == inner::LasIs) ? jmpAddr : jmpAddr - size_t(top);
 			if (jmpSize == 4) disp = inner::VerifyInInt32(disp);
 			return disp;
 		}
@@ -785,18 +1052,19 @@ class CodeArray {
 	Allocator *alloc_;
 protected:
 	size_t maxSize_;
-	uint8 *top_;
+	uint8_t *top_;
 	size_t size_;
 	bool isCalledCalcJmpAddress_;
 
+	bool useProtect() const { return alloc_->useProtect(); }
 	/*
 		allocate new memory and copy old data to the new area
 	*/
 	void growMemory()
 	{
 		const size_t newSize = (std::max<size_t>)(DEFAULT_MAX_CODE_SIZE, maxSize_ * 2);
-		uint8 *newTop = alloc_->alloc(newSize);
-		if (newTop == 0) throw Error(ERR_CANT_ALLOC);
+		uint8_t *newTop = alloc_->alloc(newSize);
+		if (newTop == 0) XBYAK_THROW(ERR_CANT_ALLOC)
 		for (size_t i = 0; i < size_; i++) newTop[i] = top_[i];
 		alloc_->free(top_);
 		top_ = newTop;
@@ -809,38 +1077,52 @@ protected:
 	{
 		if (isCalledCalcJmpAddress_) return;
 		for (AddrInfoList::const_iterator i = addrInfoList_.begin(), ie = addrInfoList_.end(); i != ie; ++i) {
-			uint64 disp = i->getVal(top_);
+			uint64_t disp = i->getVal(top_);
 			rewrite(i->codeOffset, disp, i->jmpSize);
 		}
-		if (alloc_->useProtect() && !protect(top_, size_, true)) throw Error(ERR_CANT_PROTECT);
 		isCalledCalcJmpAddress_ = true;
 	}
 public:
+	enum ProtectMode {
+		PROTECT_RW = 0, // read/write
+		PROTECT_RWE = 1, // read/write/exec
+		PROTECT_RE = 2 // read/exec
+	};
 	explicit CodeArray(size_t maxSize, void *userPtr = 0, Allocator *allocator = 0)
-		: type_(userPtr == AutoGrow ? AUTO_GROW : userPtr ? USER_BUF : ALLOC_BUF)
+		: type_(userPtr == AutoGrow ? AUTO_GROW : (userPtr == 0 || userPtr == DontSetProtectRWE) ? ALLOC_BUF : USER_BUF)
 		, alloc_(allocator ? allocator : (Allocator*)&defaultAllocator_)
 		, maxSize_(maxSize)
-		, top_(type_ == USER_BUF ? reinterpret_cast<uint8*>(userPtr) : alloc_->alloc((std::max<size_t>)(maxSize, 1)))
+		, top_(type_ == USER_BUF ? reinterpret_cast<uint8_t*>(userPtr) : alloc_->alloc((std::max<size_t>)(maxSize, 1)))
 		, size_(0)
 		, isCalledCalcJmpAddress_(false)
 	{
-		if (maxSize_ > 0 && top_ == 0) throw Error(ERR_CANT_ALLOC);
-		if ((type_ == ALLOC_BUF && alloc_->useProtect()) && !protect(top_, maxSize, true)) {
+		if (maxSize_ > 0 && top_ == 0) XBYAK_THROW(ERR_CANT_ALLOC)
+		if ((type_ == ALLOC_BUF && userPtr != DontSetProtectRWE && useProtect()) && !setProtectMode(PROTECT_RWE, false)) {
 			alloc_->free(top_);
-			throw Error(ERR_CANT_PROTECT);
+			XBYAK_THROW(ERR_CANT_PROTECT)
 		}
 	}
 	virtual ~CodeArray()
 	{
 		if (isAllocType()) {
-			if (alloc_->useProtect()) protect(top_, maxSize_, false);
+			if (useProtect()) setProtectModeRW(false);
 			alloc_->free(top_);
 		}
 	}
+	bool setProtectMode(ProtectMode mode, bool throwException = true)
+	{
+		bool isOK = protect(top_, maxSize_, mode);
+		if (isOK) return true;
+		if (throwException) XBYAK_THROW_RET(ERR_CANT_PROTECT, false)
+		return false;
+	}
+	bool setProtectModeRE(bool throwException = true) { return setProtectMode(PROTECT_RE, throwException); }
+	bool setProtectModeRW(bool throwException = true) { return setProtectMode(PROTECT_RW, throwException); }
 	void resetSize()
 	{
 		size_ = 0;
 		addrInfoList_.clear();
+		isCalledCalcJmpAddress_ = false;
 	}
 	void db(int code)
 	{
@@ -848,38 +1130,38 @@ public:
 			if (type_ == AUTO_GROW) {
 				growMemory();
 			} else {
-				throw Error(ERR_CODE_IS_TOO_BIG);
+				XBYAK_THROW(ERR_CODE_IS_TOO_BIG)
 			}
 		}
-		top_[size_++] = static_cast<uint8>(code);
+		top_[size_++] = static_cast<uint8_t>(code);
 	}
-	void db(const uint8 *code, int codeSize)
+	void db(const uint8_t *code, size_t codeSize)
 	{
-		for (int i = 0; i < codeSize; i++) db(code[i]);
+		for (size_t i = 0; i < codeSize; i++) db(code[i]);
 	}
-	void db(uint64 code, int codeSize)
+	void db(uint64_t code, size_t codeSize)
 	{
-		if (codeSize > 8) throw Error(ERR_BAD_PARAMETER);
-		for (int i = 0; i < codeSize; i++) db(static_cast<uint8>(code >> (i * 8)));
+		if (codeSize > 8) XBYAK_THROW(ERR_BAD_PARAMETER)
+		for (size_t i = 0; i < codeSize; i++) db(static_cast<uint8_t>(code >> (i * 8)));
 	}
-	void dw(uint32 code) { db(code, 2); }
-	void dd(uint32 code) { db(code, 4); }
-	void dq(uint64 code) { db(code, 8); }
-	const uint8 *getCode() const { return top_; }
+	void dw(uint32_t code) { db(code, 2); }
+	void dd(uint32_t code) { db(code, 4); }
+	void dq(uint64_t code) { db(code, 8); }
+	const uint8_t *getCode() const { return top_; }
 	template<class F>
-	const F getCode() const { return CastTo<F>(top_); }
-	const uint8 *getCurr() const { return &top_[size_]; }
+	const F getCode() const { return reinterpret_cast<F>(top_); }
+	const uint8_t *getCurr() const { return &top_[size_]; }
 	template<class F>
-	const F getCurr() const { return CastTo<F>(&top_[size_]); }
+	const F getCurr() const { return reinterpret_cast<F>(&top_[size_]); }
 	size_t getSize() const { return size_; }
 	void setSize(size_t size)
 	{
-		if (size > maxSize_) throw Error(ERR_OFFSET_IS_TOO_BIG);
+		if (size > maxSize_) XBYAK_THROW(ERR_OFFSET_IS_TOO_BIG)
 		size_ = size;
 	}
 	void dump() const
 	{
-		const uint8 *p = getCode();
+		const uint8_t *p = getCode();
 		size_t bufSize = getSize();
 		size_t remain = bufSize;
 		for (int i = 0; i < 4; i++) {
@@ -904,13 +1186,13 @@ public:
 		@param disp [in] offset from the next of jmp
 		@param size [in] write size(1, 2, 4, 8)
 	*/
-	void rewrite(size_t offset, uint64 disp, size_t size)
+	void rewrite(size_t offset, uint64_t disp, size_t size)
 	{
 		assert(offset < maxSize_);
-		if (size != 1 && size != 2 && size != 4 && size != 8) throw Error(ERR_BAD_PARAMETER);
-		uint8 *const data = top_ + offset;
+		if (size != 1 && size != 2 && size != 4 && size != 8) XBYAK_THROW(ERR_BAD_PARAMETER)
+		uint8_t *const data = top_ + offset;
 		for (size_t i = 0; i < size; i++) {
-			data[i] = static_cast<uint8>(disp >> (i * 8));
+			data[i] = static_cast<uint8_t>(disp >> (i * 8));
 		}
 	}
 	void save(size_t offset, size_t val, int size, inner::LabelMode mode)
@@ -923,19 +1205,36 @@ public:
 		change exec permission of memory
 		@param addr [in] buffer address
 		@param size [in] buffer size
-		@param canExec [in] true(enable to exec), false(disable to exec)
+		@param protectMode [in] mode(RW/RWE/RE)
 		@return true(success), false(failure)
 	*/
-	static inline bool protect(const void *addr, size_t size, bool canExec)
+	static inline bool protect(const void *addr, size_t size, int protectMode)
 	{
+#if defined(_WIN32)
+		const DWORD c_rw = PAGE_READWRITE;
+		const DWORD c_rwe = PAGE_EXECUTE_READWRITE;
+		const DWORD c_re = PAGE_EXECUTE_READ;
+		DWORD mode;
+#else
+		const int c_rw = PROT_READ | PROT_WRITE;
+		const int c_rwe = PROT_READ | PROT_WRITE | PROT_EXEC;
+		const int c_re = PROT_READ | PROT_EXEC;
+		int mode;
+#endif
+		switch (protectMode) {
+		case PROTECT_RW: mode = c_rw; break;
+		case PROTECT_RWE: mode = c_rwe; break;
+		case PROTECT_RE: mode = c_re; break;
+		default:
+			return false;
+		}
 #if defined(_WIN32)
 		DWORD oldProtect;
-		return VirtualProtect(const_cast<void*>(addr), size, canExec ? PAGE_EXECUTE_READWRITE : PAGE_READWRITE, &oldProtect) != 0;
+		return VirtualProtect(const_cast<void*>(addr), size, mode, &oldProtect) != 0;
 #elif defined(__GNUC__)
 		size_t pageSize = sysconf(_SC_PAGESIZE);
 		size_t iaddr = reinterpret_cast<size_t>(addr);
 		size_t roundAddr = iaddr & ~(pageSize - static_cast<size_t>(1));
-		int mode = PROT_READ | PROT_WRITE | (canExec ? PROT_EXEC : 0);
 		return mprotect(reinterpret_cast<void*>(roundAddr), size + (iaddr - roundAddr), mode) == 0;
 #else
 		return true;
@@ -947,9 +1246,9 @@ public:
 		@param alignedSize [in] power of two
 		@return aligned addr by alingedSize
 	*/
-	static inline uint8 *getAlignedAddress(uint8 *addr, size_t alignedSize = 16)
+	static inline uint8_t *getAlignedAddress(uint8_t *addr, size_t alignedSize = 16)
 	{
-		return reinterpret_cast<uint8*>((reinterpret_cast<size_t>(addr) + alignedSize - 1) & ~(alignedSize - static_cast<size_t>(1)));
+		return reinterpret_cast<uint8_t*>((reinterpret_cast<size_t>(addr) + alignedSize - 1) & ~(alignedSize - static_cast<size_t>(1)));
 	}
 };
 
@@ -961,52 +1260,55 @@ public:
 		M_rip,
 		M_ripAddr
 	};
-	Address(uint32 sizeBit, bool broadcast, const RegExp& e)
-		: Operand(0, MEM, sizeBit), e_(e), label_(0), mode_(M_ModRM), permitVsib_(false), broadcast_(broadcast)
+	XBYAK_CONSTEXPR Address(uint32_t sizeBit, bool broadcast, const RegExp& e)
+		: Operand(0, MEM, sizeBit), e_(e), label_(0), mode_(M_ModRM), broadcast_(broadcast)
 	{
 		e_.verify();
-		e_.optimize();
 	}
 #ifdef XBYAK64
-	explicit Address(size_t disp)
-		: Operand(0, MEM, 64), e_(disp), label_(0), mode_(M_64bitDisp), permitVsib_(false), broadcast_(false){ }
-	Address(uint32 sizeBit, bool broadcast, const RegRip& addr)
-		: Operand(0, MEM, sizeBit), e_(addr.disp_), label_(addr.label_), mode_(addr.isAddr_ ? M_ripAddr : M_rip), permitVsib_(false), broadcast_(broadcast) { }
+	explicit XBYAK_CONSTEXPR Address(size_t disp)
+		: Operand(0, MEM, 64), e_(disp), label_(0), mode_(M_64bitDisp), broadcast_(false){ }
+	XBYAK_CONSTEXPR Address(uint32_t sizeBit, bool broadcast, const RegRip& addr)
+		: Operand(0, MEM, sizeBit), e_(addr.disp_), label_(addr.label_), mode_(addr.isAddr_ ? M_ripAddr : M_rip), broadcast_(broadcast) { }
 #endif
-	void permitVsib() const { permitVsib_ = true; }
-	const RegExp& getRegExp() const { return e_; }
+	RegExp getRegExp(bool optimize = true) const
+	{
+		return optimize ? e_.optimize() : e_;
+	}
 	Mode getMode() const { return mode_; }
-	bool is32bit() const { verify(); return e_.getBase().getBit() == 32 || e_.getIndex().getBit() == 32; }
-	bool isOnlyDisp() const { verify(); return !e_.getBase().getBit() && !e_.getIndex().getBit(); } // for mov eax
-	size_t getDisp() const { verify(); return e_.getDisp(); }
-	uint8 getRex() const
+	bool is32bit() const { return e_.getBase().getBit() == 32 || e_.getIndex().getBit() == 32; }
+	bool isOnlyDisp() const { return !e_.getBase().getBit() && !e_.getIndex().getBit(); } // for mov eax
+	size_t getDisp() const { return e_.getDisp(); }
+	uint8_t getRex() const
 	{
-		verify();
 		if (mode_ != M_ModRM) return 0;
-		uint8 rex = e_.getIndex().getRexX() | e_.getBase().getRexB();
-		if (rex) rex |= 0x40;
-		return rex;
+		return getRegExp().getRex();
 	}
-	bool is64bitDisp() const { verify(); return mode_ == M_64bitDisp; } // for moffset
+	bool is64bitDisp() const { return mode_ == M_64bitDisp; } // for moffset
 	bool isBroadcast() const { return broadcast_; }
 	const Label* getLabel() const { return label_; }
 	bool operator==(const Address& rhs) const
 	{
-		return getBit() == rhs.getBit() && e_ == rhs.e_ && label_ == rhs.label_ && mode_ == rhs.mode_ && permitVsib_ == rhs.permitVsib_ && broadcast_ == rhs.broadcast_;
+		return getBit() == rhs.getBit() && e_ == rhs.e_ && label_ == rhs.label_ && mode_ == rhs.mode_ && broadcast_ == rhs.broadcast_;
 	}
 	bool operator!=(const Address& rhs) const { return !operator==(rhs); }
+	bool isVsib() const { return e_.isVsib(); }
 private:
 	RegExp e_;
 	const Label* label_;
 	Mode mode_;
-	mutable bool permitVsib_;
 	bool broadcast_;
-	void verify() const { if (e_.isVsib() && !permitVsib_) throw Error(ERR_BAD_VSIB_ADDRESSING); }
 };
 
+inline const Address& Operand::getAddress() const
+{
+	assert(isMEM());
+	return static_cast<const Address&>(*this);
+}
+
 inline bool Operand::operator==(const Operand& rhs) const
 {
-	if (isMEM() && rhs.isMEM()) return static_cast<const Address&>(*this) == static_cast<const Address&>(rhs);
+	if (isMEM() && rhs.isMEM()) return this->getAddress() == rhs.getAddress();
 	return isEqualIfNotInherited(rhs);
 }
 
@@ -1014,9 +1316,9 @@ class AddressFrame {
 	void operator=(const AddressFrame&);
 	AddressFrame(const AddressFrame&);
 public:
-	const uint32 bit_;
+	const uint32_t bit_;
 	const bool broadcast_;
-	explicit AddressFrame(uint32 bit, bool broadcast = false) : bit_(bit), broadcast_(broadcast) { }
+	explicit XBYAK_CONSTEXPR AddressFrame(uint32_t bit, bool broadcast = false) : bit_(bit), broadcast_(broadcast) { }
 	Address operator[](const RegExp& e) const
 	{
 		return Address(bit_, broadcast_, e);
@@ -1026,7 +1328,7 @@ public:
 		return Address(bit_, broadcast_, RegExp(reinterpret_cast<size_t>(disp)));
 	}
 #ifdef XBYAK64
-	Address operator[](uint64 disp) const { return Address(disp); }
+	Address operator[](uint64_t disp) const { return Address(disp); }
 	Address operator[](const RegRip& addr) const { return Address(bit_, broadcast_, addr); }
 #endif
 };
@@ -1053,8 +1355,9 @@ public:
 	Label(const Label& rhs);
 	Label& operator=(const Label& rhs);
 	~Label();
+	void clear() { mgr = 0; id = 0; }
 	int getId() const { return id; }
-	const uint8 *getAddress() const;
+	const uint8_t *getAddress() const;
 
 	// backward compatibility
 	static inline std::string toStr(int num)
@@ -1091,6 +1394,7 @@ class LabelManager {
 	};
 	typedef XBYAK_STD_UNORDERED_MAP<int, ClabelVal> ClabelDefList;
 	typedef XBYAK_STD_UNORDERED_MULTIMAP<int, const JmpLabel> ClabelUndefList;
+	typedef XBYAK_STD_UNORDERED_SET<Label*> LabelPtrList;
 
 	CodeArray *base_;
 	// global : stateList_.front(), local : stateList_.back()
@@ -1098,6 +1402,7 @@ class LabelManager {
 	mutable int labelId_;
 	ClabelDefList clabelDefList_;
 	ClabelUndefList clabelUndefList_;
+	LabelPtrList labelPtrList_;
 
 	int getId(const Label& label) const
 	{
@@ -1110,7 +1415,7 @@ class LabelManager {
 		// add label
 		typename DefList::value_type item(labelId, addrOffset);
 		std::pair<typename DefList::iterator, bool> ret = defList.insert(item);
-		if (!ret.second) throw Error(ERR_LABEL_IS_REDEFINED);
+		if (!ret.second) XBYAK_THROW(ERR_LABEL_IS_REDEFINED)
 		// search undefined label
 		for (;;) {
 			typename UndefList::iterator itr = undefList.find(labelId);
@@ -1125,9 +1430,9 @@ class LabelManager {
 			} else {
 				disp = addrOffset - jmp->endOfJmp + jmp->disp;
 #ifdef XBYAK64
-				if (jmp->jmpSize <= 4 && !inner::IsInInt32(disp)) throw Error(ERR_OFFSET_IS_TOO_BIG);
+				if (jmp->jmpSize <= 4 && !inner::IsInInt32(disp)) XBYAK_THROW(ERR_OFFSET_IS_TOO_BIG)
 #endif
-				if (jmp->jmpSize == 1 && !inner::IsInDisp8((uint32)disp)) throw Error(ERR_LABEL_IS_TOO_FAR);
+				if (jmp->jmpSize == 1 && !inner::IsInDisp8((uint32_t)disp)) XBYAK_THROW(ERR_LABEL_IS_TOO_FAR)
 			}
 			if (base_->isAutoGrow()) {
 				base_->save(offset, disp, jmp->jmpSize, jmp->mode);
@@ -1146,9 +1451,14 @@ class LabelManager {
 		return true;
 	}
 	friend class Label;
-	void incRefCount(int id) { clabelDefList_[id].refCount++; }
-	void decRefCount(int id)
+	void incRefCount(int id, Label *label)
 	{
+		clabelDefList_[id].refCount++;
+		labelPtrList_.insert(label);
+	}
+	void decRefCount(int id, Label *label)
+	{
+		labelPtrList_.erase(label);
 		ClabelDefList::iterator i = clabelDefList_.find(id);
 		if (i == clabelDefList_.end()) return;
 		if (i->second.refCount == 1) {
@@ -1167,11 +1477,23 @@ class LabelManager {
 #endif
 		return !list.empty();
 	}
+	// detach all labels linked to LabelManager
+	void resetLabelPtrList()
+	{
+		for (LabelPtrList::iterator i = labelPtrList_.begin(), ie = labelPtrList_.end(); i != ie; ++i) {
+			(*i)->clear();
+		}
+		labelPtrList_.clear();
+	}
 public:
 	LabelManager()
 	{
 		reset();
 	}
+	~LabelManager()
+	{
+		resetLabelPtrList();
+	}
 	void reset()
 	{
 		base_ = 0;
@@ -1181,6 +1503,7 @@ public:
 		stateList_.push_back(SlabelState());
 		clabelDefList_.clear();
 		clabelUndefList_.clear();
+		resetLabelPtrList();
 	}
 	void enterLocal()
 	{
@@ -1188,14 +1511,14 @@ public:
 	}
 	void leaveLocal()
 	{
-		if (stateList_.size() <= 2) throw Error(ERR_UNDER_LOCAL_LABEL);
-		if (hasUndefinedLabel_inner(stateList_.back().undefList)) throw Error(ERR_LABEL_IS_NOT_FOUND);
+		if (stateList_.size() <= 2) XBYAK_THROW(ERR_UNDER_LOCAL_LABEL)
+		if (hasUndefinedLabel_inner(stateList_.back().undefList)) XBYAK_THROW(ERR_LABEL_IS_NOT_FOUND)
 		stateList_.pop_back();
 	}
 	void set(CodeArray *base) { base_ = base; }
 	void defineSlabel(std::string label)
 	{
-		if (label == "@b" || label == "@f") throw Error(ERR_BAD_LABEL_STR);
+		if (label == "@b" || label == "@f") XBYAK_THROW(ERR_BAD_LABEL_STR)
 		if (label == "@@") {
 			SlabelDefList& defList = stateList_.front().defList;
 			SlabelDefList::iterator i = defList.find("@f");
@@ -1213,17 +1536,19 @@ public:
 		SlabelState& st = *label.c_str() == '.' ? stateList_.back() : stateList_.front();
 		define_inner(st.defList, st.undefList, label, base_->getSize());
 	}
-	void defineClabel(const Label& label)
+	void defineClabel(Label& label)
 	{
 		define_inner(clabelDefList_, clabelUndefList_, getId(label), base_->getSize());
 		label.mgr = this;
+		labelPtrList_.insert(&label);
 	}
 	void assign(Label& dst, const Label& src)
 	{
 		ClabelDefList::const_iterator i = clabelDefList_.find(src.id);
-		if (i == clabelDefList_.end()) throw Error(ERR_LABEL_ISNOT_SET_BY_L);
+		if (i == clabelDefList_.end()) XBYAK_THROW(ERR_LABEL_ISNOT_SET_BY_L)
 		define_inner(clabelDefList_, clabelUndefList_, dst.id, i->second.offset);
 		dst.mgr = this;
+		labelPtrList_.insert(&dst);
 	}
 	bool getOffset(size_t *offset, std::string& label) const
 	{
@@ -1232,7 +1557,7 @@ public:
 			if (defList.find("@f") != defList.end()) {
 				label = "@f";
 			} else if (defList.find("@b") == defList.end()) {
-				throw Error(ERR_LABEL_IS_NOT_FOUND);
+				XBYAK_THROW_RET(ERR_LABEL_IS_NOT_FOUND, false)
 			}
 		} else if (label == "@f") {
 			if (defList.find("@f") != defList.end()) {
@@ -1263,7 +1588,7 @@ public:
 		return false;
 	}
 	bool hasUndefClabel() const { return hasUndefinedLabel_inner(clabelUndefList_); }
-	const uint8 *getCode() const { return base_->getCode(); }
+	const uint8_t *getCode() const { return base_->getCode(); }
 	bool isReady() const { return !base_->isAutoGrow() || base_->isCalledCalcJmpAddress(); }
 };
 
@@ -1271,21 +1596,21 @@ inline Label::Label(const Label& rhs)
 {
 	id = rhs.id;
 	mgr = rhs.mgr;
-	if (mgr) mgr->incRefCount(id);
+	if (mgr) mgr->incRefCount(id, this);
 }
 inline Label& Label::operator=(const Label& rhs)
 {
-	if (id) throw Error(ERR_LABEL_IS_ALREADY_SET_BY_L);
+	if (id) XBYAK_THROW_RET(ERR_LABEL_IS_ALREADY_SET_BY_L, *this)
 	id = rhs.id;
 	mgr = rhs.mgr;
-	if (mgr) mgr->incRefCount(id);
+	if (mgr) mgr->incRefCount(id, this);
 	return *this;
 }
 inline Label::~Label()
 {
-	if (id && mgr) mgr->decRefCount(id);
+	if (id && mgr) mgr->decRefCount(id, this);
 }
-inline const uint8* Label::getAddress() const
+inline const uint8_t* Label::getAddress() const
 {
 	if (mgr == 0 || !mgr->isReady()) return 0;
 	size_t offset;
@@ -1293,18 +1618,25 @@ inline const uint8* Label::getAddress() const
 	return mgr->getCode() + offset;
 }
 
+typedef enum {
+	DefaultEncoding,
+	VexEncoding,
+	EvexEncoding
+} PreferredEncoding;
+
 class CodeGenerator : public CodeArray {
 public:
 	enum LabelType {
 		T_SHORT,
 		T_NEAR,
+		T_FAR, // far jump
 		T_AUTO // T_SHORT if possible
 	};
 private:
 	CodeGenerator operator=(const CodeGenerator&); // don't call
 #ifdef XBYAK64
 	enum { i32e = 32 | 64, BIT = 64 };
-	static const size_t dummyAddr = (size_t(0x11223344) << 32) | 55667788;
+	static const uint64_t dummyAddr = uint64_t(0x1122334455667788ull);
 	typedef Reg64 NativeReg;
 #else
 	enum { i32e = 32, BIT = 32 };
@@ -1346,59 +1678,76 @@ private:
 	{
 		return op1.isREG(i32e) && ((op2.isREG(i32e) && op1.getBit() == op2.getBit()) || op2.isMEM());
 	}
+	static inline bool isValidSSE(const Operand& op1)
+	{
+		// SSE instructions do not support XMM16 - XMM31
+		return !(op1.isXMM() && op1.getIdx() >= 16);
+	}
 	void rex(const Operand& op1, const Operand& op2 = Operand())
 	{
-		uint8 rex = 0;
+		uint8_t rex = 0;
 		const Operand *p1 = &op1, *p2 = &op2;
 		if (p1->isMEM()) std::swap(p1, p2);
-		if (p1->isMEM()) throw Error(ERR_BAD_COMBINATION);
+		if (p1->isMEM()) XBYAK_THROW(ERR_BAD_COMBINATION)
 		if (p2->isMEM()) {
-			const Address& addr = static_cast<const Address&>(*p2);
+			const Address& addr = p2->getAddress();
 			if (BIT == 64 && addr.is32bit()) db(0x67);
-			rex = addr.getRex() | static_cast<const Reg&>(*p1).getRex();
+			rex = addr.getRex() | p1->getReg().getRex();
 		} else {
 			// ModRM(reg, base);
-			rex = static_cast<const Reg&>(op2).getRex(static_cast<const Reg&>(op1));
+			rex = op2.getReg().getRex(op1.getReg());
 		}
 		// except movsx(16bit, 32/64bit)
 		if ((op1.isBit(16) && !op2.isBit(i32e)) || (op2.isBit(16) && !op1.isBit(i32e))) db(0x66);
 		if (rex) db(rex);
 	}
 	enum AVXtype {
-		T_66 = 1 << 0,
-		T_F3 = 1 << 1,
-		T_F2 = 1 << 2,
-		T_0F = 1 << 3,
-		T_0F38 = 1 << 4,
-		T_0F3A = 1 << 5,
-		T_L0 = 1 << 6,
-		T_L1 = 1 << 7,
-		T_W0 = 1 << 8,
-		T_W1 = 1 << 9,
-		T_EW0 = 1 << 10,
-		T_EW1 = 1 << 11,
-		T_YMM = 1 << 12, // support YMM, ZMM
-		T_EVEX = 1 << 13,
-		T_ER_X = 1 << 14, // xmm{er}
-		T_ER_Y = 1 << 15, // ymm{er}
-		T_ER_Z = 1 << 16, // zmm{er}
-		T_SAE_X = 1 << 17, // xmm{sae}
-		T_SAE_Y = 1 << 18, // ymm{sae}
-		T_SAE_Z = 1 << 19, // zmm{sae}
-		T_MUST_EVEX = 1 << 20, // contains T_EVEX
-		T_B32 = 1 << 21, // m32bcst
-		T_B64 = 1 << 22, // m64bcst
-		T_M_K = 1 << 23, // mem{k}
-		T_N1 = 1 << 24,
-		T_N2 = 1 << 25,
-		T_N4 = 1 << 26,
-		T_N8 = 1 << 27,
-		T_N16 = 1 << 28,
-		T_N32 = 1 << 29,
-		T_N_VL = 1 << 30, // N * (1, 2, 4) for VL
-		T_DUP = 1 << 31, // N = (8, 32, 64)
+		// low 3 bit
+		T_N1 = 1,
+		T_N2 = 2,
+		T_N4 = 3,
+		T_N8 = 4,
+		T_N16 = 5,
+		T_N32 = 6,
+		T_NX_MASK = 7,
+		//
+		T_N_VL = 1 << 3, // N * (1, 2, 4) for VL
+		T_DUP = 1 << 4, // N = (8, 32, 64)
+		T_66 = 1 << 5, // pp = 1
+		T_F3 = 1 << 6, // pp = 2
+		T_F2 = T_66 | T_F3, // pp = 3
+		T_ER_R = 1 << 7, // reg{er}
+		T_0F = 1 << 8,
+		T_0F38 = 1 << 9,
+		T_0F3A = 1 << 10,
+		T_L0 = 1 << 11,
+		T_L1 = 1 << 12,
+		T_W0 = 1 << 13,
+		T_W1 = 1 << 14,
+		T_EW0 = 1 << 15,
+		T_EW1 = 1 << 16,
+		T_YMM = 1 << 17, // support YMM, ZMM
+		T_EVEX = 1 << 18,
+		T_ER_X = 1 << 19, // xmm{er}
+		T_ER_Y = 1 << 20, // ymm{er}
+		T_ER_Z = 1 << 21, // zmm{er}
+		T_SAE_X = 1 << 22, // xmm{sae}
+		T_SAE_Y = 1 << 23, // ymm{sae}
+		T_SAE_Z = 1 << 24, // zmm{sae}
+		T_MUST_EVEX = 1 << 25, // contains T_EVEX
+		T_B32 = 1 << 26, // m32bcst
+		T_B64 = 1 << 27, // m64bcst
+		T_B16 = T_B32 | T_B64, // m16bcst (Be careful)
+		T_M_K = 1 << 28, // mem{k}
+		T_VSIB = 1 << 29,
+		T_MEM_EVEX = 1 << 30, // use evex if mem
+		T_FP16 = 1 << 31, // avx512-fp16
+		T_MAP5 = T_FP16 | T_0F,
+		T_MAP6 = T_FP16 | T_0F38,
 		T_XXX
 	};
+	// T_66 = 1, T_F3 = 2, T_F2 = 3
+	uint32_t getPP(int type) const { return (type >> 5) & 3; }
 	void vex(const Reg& reg, const Reg& base, const Operand *v, int type, int code, bool x = false)
 	{
 		int w = (type & T_W1) ? 1 : 0;
@@ -1406,13 +1755,13 @@ private:
 		bool r = reg.isExtIdx();
 		bool b = base.isExtIdx();
 		int idx = v ? v->getIdx() : 0;
-		if ((idx | reg.getIdx() | base.getIdx()) >= 16) throw Error(ERR_BAD_COMBINATION);
-		uint32 pp = (type & T_66) ? 1 : (type & T_F3) ? 2 : (type & T_F2) ? 3 : 0;
-		uint32 vvvv = (((~idx) & 15) << 3) | (is256 ? 4 : 0) | pp;
+		if ((idx | reg.getIdx() | base.getIdx()) >= 16) XBYAK_THROW(ERR_BAD_COMBINATION)
+		uint32_t pp = getPP(type);
+		uint32_t vvvv = (((~idx) & 15) << 3) | (is256 ? 4 : 0) | pp;
 		if (!b && !x && !w && (type & T_0F)) {
 			db(0xC5); db((r ? 0 : 0x80) | vvvv);
 		} else {
-			uint32 mmmm = (type & T_0F) ? 1 : (type & T_0F38) ? 2 : (type & T_0F3A) ? 3 : 0;
+			uint32_t mmmm = (type & T_0F) ? 1 : (type & T_0F38) ? 2 : (type & T_0F3A) ? 3 : 0;
 			db(0xC4); db((r ? 0 : 0x80) | (x ? 0 : 0x40) | (b ? 0 : 0x20) | mmmm); db((w << 7) | vvvv);
 		}
 		db(code);
@@ -1420,36 +1769,30 @@ private:
 	void verifySAE(const Reg& r, int type) const
 	{
 		if (((type & T_SAE_X) && r.isXMM()) || ((type & T_SAE_Y) && r.isYMM()) || ((type & T_SAE_Z) && r.isZMM())) return;
-		throw Error(ERR_SAE_IS_INVALID);
+		XBYAK_THROW(ERR_SAE_IS_INVALID)
 	}
 	void verifyER(const Reg& r, int type) const
 	{
+		if ((type & T_ER_R) && r.isREG(32|64)) return;
 		if (((type & T_ER_X) && r.isXMM()) || ((type & T_ER_Y) && r.isYMM()) || ((type & T_ER_Z) && r.isZMM())) return;
-		throw Error(ERR_ER_IS_INVALID);
+		XBYAK_THROW(ERR_ER_IS_INVALID)
 	}
 	// (a, b, c) contains non zero two or three values then err
 	int verifyDuplicate(int a, int b, int c, int err)
 	{
 		int v = a | b | c;
-		if ((a > 0 && a != v) + (b > 0 && b != v) + (c > 0 && c != v) > 0) return Error(err);
+		if ((a > 0 && a != v) + (b > 0 && b != v) + (c > 0 && c != v) > 0) XBYAK_THROW_RET(err, 0)
 		return v;
 	}
-	enum {
-		T_RN_SAE = 1,
-		T_RD_SAE = 2,
-		T_RU_SAE = 3,
-		T_RZ_SAE = 4,
-		T_SAE = 5
-	};
-	int evex(const Reg& reg, const Reg& base, const Operand *v, int type, int code, bool x = false, bool b = false, int aaa = 0, uint32 VL = 0)
+	int evex(const Reg& reg, const Reg& base, const Operand *v, int type, int code, bool x = false, bool b = false, int aaa = 0, uint32_t VL = 0, bool Hi16Vidx = false)
 	{
-		if (!(type & (T_EVEX | T_MUST_EVEX))) throw Error(ERR_EVEX_IS_INVALID);
+		if (!(type & (T_EVEX | T_MUST_EVEX))) XBYAK_THROW_RET(ERR_EVEX_IS_INVALID, 0)
 		int w = (type & T_EW1) ? 1 : 0;
-		uint32 mm = (type & T_0F) ? 1 : (type & T_0F38) ? 2 : (type & T_0F3A) ? 3 : 0;
-		uint32 pp = (type & T_66) ? 1 : (type & T_F3) ? 2 : (type & T_F2) ? 3 : 0;
-
+		uint32_t mmm = (type & T_0F) ? 1 : (type & T_0F38) ? 2 : (type & T_0F3A) ? 3 : 0;
+		if (type & T_FP16) mmm |= 4;
+		uint32_t pp = getPP(type);
 		int idx = v ? v->getIdx() : 0;
-		uint32 vvvv = ~idx;
+		uint32_t vvvv = ~idx;
 
 		bool R = !reg.isExtIdx();
 		bool X = x ? false : !base.isExtIdx2();
@@ -1459,7 +1802,7 @@ private:
 		int rounding = verifyDuplicate(reg.getRounding(), base.getRounding(), v ? v->getRounding() : 0, ERR_ROUNDING_IS_ALREADY_SET);
 		int disp8N = 1;
 		if (rounding) {
-			if (rounding == T_SAE){
+			if (rounding == EvexModifierRounding::T_SAE) {
 				verifySAE(base, type); LL = 0;
 			} else {
 				verifyER(base, type); LL = rounding - 1;
@@ -1470,24 +1813,26 @@ private:
 			VL = (std::max)((std::max)(reg.getBit(), base.getBit()), VL);
 			LL = (VL == 512) ? 2 : (VL == 256) ? 1 : 0;
 			if (b) {
-				disp8N = (type & T_B32) ? 4 : 8;
+				disp8N = ((type & T_B16) == T_B16) ? 2 : (type & T_B32) ? 4 : 8;
 			} else if (type & T_DUP) {
 				disp8N = VL == 128 ? 8 : VL == 256 ? 32 : 64;
 			} else {
-				if ((type & (T_N1 | T_N2 | T_N4 | T_N8 | T_N16 | T_N32 | T_N_VL)) == 0) {
+				if ((type & (T_NX_MASK | T_N_VL)) == 0) {
 					type |= T_N16 | T_N_VL; // default
 				}
-				if (type & (T_N1 | T_N2 | T_N4 | T_N8 | T_N16 | T_N32)) {
-					disp8N = (type & T_N1) ? 1 : (type & T_N2) ? 2 : (type & T_N4) ? 4 : (type & T_N8) ? 8 : (type & T_N16) ? 16 : 32;
+				int low = type & T_NX_MASK;
+				if (low > 0) {
+					disp8N = 1 << (low - 1);
 					if (type & T_N_VL) disp8N *= (VL == 512 ? 4 : VL == 256 ? 2 : 1);
 				}
 			}
 		}
-		bool Vp = !(v ? v->isExtIdx2() : 0);
+		bool Vp = !((v ? v->isExtIdx2() : 0) | Hi16Vidx);
 		bool z = reg.hasZero() || base.hasZero() || (v ? v->hasZero() : false);
 		if (aaa == 0) aaa = verifyDuplicate(base.getOpmaskIdx(), reg.getOpmaskIdx(), (v ? v->getOpmaskIdx() : 0), ERR_OPMASK_IS_ALREADY_SET);
+		if (aaa == 0) z = 0; // clear T_z if mask is not set
 		db(0x62);
-		db((R ? 0x80 : 0) | (X ? 0x40 : 0) | (B ? 0x20 : 0) | (Rp ? 0x10 : 0) | (mm & 3));
+		db((R ? 0x80 : 0) | (X ? 0x40 : 0) | (B ? 0x20 : 0) | (Rp ? 0x10 : 0) | mmm);
 		db((w == 1 ? 0x80 : 0) | ((vvvv & 15) << 3) | 4 | (pp & 3));
 		db((z ? 0x80 : 0) | ((LL & 3) << 5) | (b ? 0x10 : 0) | (Vp ? 8 : 0) | (aaa & 7));
 		db(code);
@@ -1495,16 +1840,23 @@ private:
 	}
 	void setModRM(int mod, int r1, int r2)
 	{
-		db(static_cast<uint8>((mod << 6) | ((r1 & 7) << 3) | (r2 & 7)));
+		db(static_cast<uint8_t>((mod << 6) | ((r1 & 7) << 3) | (r2 & 7)));
 	}
 	void setSIB(const RegExp& e, int reg, int disp8N = 0)
 	{
-		size_t disp64 = e.getDisp();
-#ifdef XBYAK64
-		size_t high = disp64 >> 32;
-		if (high != 0 && high != 0xFFFFFFFF) throw Error(ERR_OFFSET_IS_TOO_BIG);
+		uint64_t disp64 = e.getDisp();
+#if defined(XBYAK64) && !defined(__ILP32__)
+#ifdef XBYAK_OLD_DISP_CHECK
+		// treat 0xffffffff as 0xffffffffffffffff
+		uint64_t high = disp64 >> 32;
+		if (high != 0 && high != 0xFFFFFFFF) XBYAK_THROW(ERR_OFFSET_IS_TOO_BIG)
+#else
+		// displacement should be a signed 32-bit value, so also check sign bit
+		uint64_t high = disp64 >> 31;
+		if (high != 0 && high != 0x1FFFFFFFF) XBYAK_THROW(ERR_OFFSET_IS_TOO_BIG)
+#endif
 #endif
-		uint32 disp = static_cast<uint32>(disp64);
+		uint32_t disp = static_cast<uint32_t>(disp64);
 		const Reg& base = e.getBase();
 		const Reg& index = e.getIndex();
 		const int baseIdx = base.getIdx();
@@ -1523,7 +1875,7 @@ private:
 				}
 			} else {
 				// disp must be casted to signed
-				uint32 t = static_cast<uint32>(static_cast<int>(disp) / disp8N);
+				uint32_t t = static_cast<uint32_t>(static_cast<int>(disp) / disp8N);
 				if ((disp % disp8N) == 0 && inner::IsInDisp8(t)) {
 					disp = t;
 					mod = mod01;
@@ -1553,7 +1905,7 @@ private:
 		}
 	}
 	LabelManager labelMgr_;
-	bool isInDisp16(uint32 x) const { return 0xFFFF8000 <= x || x <= 0xFFFF; }
+	bool isInDisp16(uint32_t x) const { return 0xFFFF8000 <= x || x <= 0x7FFF; }
 	void opModR(const Reg& reg1, const Reg& reg2, int code0, int code1 = NONE, int code2 = NONE)
 	{
 		rex(reg2, reg1);
@@ -1562,12 +1914,31 @@ private:
 	}
 	void opModM(const Address& addr, const Reg& reg, int code0, int code1 = NONE, int code2 = NONE, int immSize = 0)
 	{
-		if (addr.is64bitDisp()) throw Error(ERR_CANT_USE_64BIT_DISP);
+		if (addr.is64bitDisp()) XBYAK_THROW(ERR_CANT_USE_64BIT_DISP)
 		rex(addr, reg);
 		db(code0 | (reg.isBit(8) ? 0 : 1)); if (code1 != NONE) db(code1); if (code2 != NONE) db(code2);
 		opAddr(addr, reg.getIdx(), immSize);
 	}
-	void makeJmp(uint32 disp, LabelType type, uint8 shortCode, uint8 longCode, uint8 longPref)
+	void opLoadSeg(const Address& addr, const Reg& reg, int code0, int code1 = NONE)
+	{
+		if (addr.is64bitDisp()) XBYAK_THROW(ERR_CANT_USE_64BIT_DISP)
+		if (reg.isBit(8)) XBYAK_THROW(ERR_BAD_SIZE_OF_REGISTER)
+		rex(addr, reg);
+		db(code0); if (code1 != NONE) db(code1);
+		opAddr(addr, reg.getIdx());
+	}
+	void opMIB(const Address& addr, const Reg& reg, int code0, int code1)
+	{
+		if (addr.is64bitDisp()) XBYAK_THROW(ERR_CANT_USE_64BIT_DISP)
+		if (addr.getMode() != Address::M_ModRM) XBYAK_THROW(ERR_INVALID_MIB_ADDRESS)
+		if (BIT == 64 && addr.is32bit()) db(0x67);
+		const RegExp& regExp = addr.getRegExp(false);
+		uint8_t rex = regExp.getRex();
+		if (rex) db(rex);
+		db(code0); db(code1);
+		setSIB(regExp, reg.getIdx());
+	}
+	void makeJmp(uint32_t disp, LabelType type, uint8_t shortCode, uint8_t longCode, uint8_t longPref)
 	{
 		const int shortJmpSize = 2;
 		const int longHeaderSize = longPref ? 2 : 1;
@@ -1575,21 +1946,23 @@ private:
 		if (type != T_NEAR && inner::IsInDisp8(disp - shortJmpSize)) {
 			db(shortCode); db(disp - shortJmpSize);
 		} else {
-			if (type == T_SHORT) throw Error(ERR_LABEL_IS_TOO_FAR);
+			if (type == T_SHORT) XBYAK_THROW(ERR_LABEL_IS_TOO_FAR)
 			if (longPref) db(longPref);
 			db(longCode); dd(disp - longJmpSize);
 		}
 	}
+	bool isNEAR(LabelType type) const { return type == T_NEAR || (type == T_AUTO && isDefaultJmpNEAR_); }
 	template<class T>
-	void opJmp(T& label, LabelType type, uint8 shortCode, uint8 longCode, uint8 longPref)
+	void opJmp(T& label, LabelType type, uint8_t shortCode, uint8_t longCode, uint8_t longPref)
 	{
+		if (type == T_FAR) XBYAK_THROW(ERR_NOT_SUPPORTED)
 		if (isAutoGrow() && size_ + 16 >= maxSize_) growMemory(); /* avoid splitting code of jmp */
 		size_t offset = 0;
 		if (labelMgr_.getOffset(&offset, label)) { /* label exists */
 			makeJmp(inner::VerifyInInt32(offset - size_), type, shortCode, longCode, longPref);
 		} else {
 			int jmpSize = 0;
-			if (type == T_NEAR) {
+			if (isNEAR(type)) {
 				jmpSize = 4;
 				if (longPref) db(longPref);
 				db(longCode); dd(0);
@@ -1601,25 +1974,37 @@ private:
 			labelMgr_.addUndefinedLabel(label, jmp);
 		}
 	}
-	void opJmpAbs(const void *addr, LabelType type, uint8 shortCode, uint8 longCode, uint8 longPref = 0)
+	void opJmpAbs(const void *addr, LabelType type, uint8_t shortCode, uint8_t longCode, uint8_t longPref = 0)
 	{
+		if (type == T_FAR) XBYAK_THROW(ERR_NOT_SUPPORTED)
 		if (isAutoGrow()) {
-			if (type != T_NEAR) throw Error(ERR_ONLY_T_NEAR_IS_SUPPORTED_IN_AUTO_GROW);
+			if (!isNEAR(type)) XBYAK_THROW(ERR_ONLY_T_NEAR_IS_SUPPORTED_IN_AUTO_GROW)
 			if (size_ + 16 >= maxSize_) growMemory();
 			if (longPref) db(longPref);
 			db(longCode);
 			dd(0);
 			save(size_ - 4, size_t(addr) - size_, 4, inner::Labs);
 		} else {
-			makeJmp(inner::VerifyInInt32(reinterpret_cast<const uint8*>(addr) - getCurr()), type, shortCode, longCode, longPref);
+			makeJmp(inner::VerifyInInt32(reinterpret_cast<const uint8_t*>(addr) - getCurr()), type, shortCode, longCode, longPref);
 		}
 
 	}
+	void opJmpOp(const Operand& op, LabelType type, int ext)
+	{
+		const int bit = 16|i32e;
+		if (type == T_FAR) {
+			if (!op.isMEM(bit)) XBYAK_THROW(ERR_NOT_SUPPORTED)
+			opR_ModM(op, bit, ext + 1, 0xFF, NONE, NONE, false);
+		} else {
+			opR_ModM(op, bit, ext, 0xFF, NONE, NONE, true);
+		}
+	}
 	// reg is reg field of ModRM
 	// immSize is the size for immediate value
 	// disp8N = 0(normal), disp8N = 1(force disp32), disp8N = {2, 4, 8} ; compressed displacement
-	void opAddr(const Address &addr, int reg, int immSize = 0, int disp8N = 0)
+	void opAddr(const Address &addr, int reg, int immSize = 0, int disp8N = 0, bool permitVisb = false)
 	{
+		if (!permitVisb && addr.isVsib()) XBYAK_THROW(ERR_BAD_VSIB_ADDRESSING)
 		if (addr.getMode() == Address::M_ModRM) {
 			setSIB(addr.getRegExp(), reg, disp8N);
 		} else if (addr.getMode() == Address::M_rip || addr.getMode() == Address::M_ripAddr) {
@@ -1629,7 +2014,7 @@ private:
 			} else {
 				size_t disp = addr.getDisp();
 				if (addr.getMode() == Address::M_ripAddr) {
-					if (isAutoGrow()) throw Error(ERR_INVALID_RIP_IN_AUTO_GROW);
+					if (isAutoGrow()) XBYAK_THROW(ERR_INVALID_RIP_IN_AUTO_GROW)
 					disp -= (size_t)getCurr() + 4 + immSize;
 				}
 				dd(inner::VerifyInInt32(disp));
@@ -1639,17 +2024,19 @@ private:
 	/* preCode is for SSSE3/SSE4 */
 	void opGen(const Operand& reg, const Operand& op, int code, int pref, bool isValid(const Operand&, const Operand&), int imm8 = NONE, int preCode = NONE)
 	{
-		if (isValid && !isValid(reg, op)) throw Error(ERR_BAD_COMBINATION);
+		if (isValid && !isValid(reg, op)) XBYAK_THROW(ERR_BAD_COMBINATION)
+		if (!isValidSSE(reg) || !isValidSSE(op)) XBYAK_THROW(ERR_NOT_SUPPORTED)
 		if (pref != NONE) db(pref);
 		if (op.isMEM()) {
-			opModM(static_cast<const Address&>(op), static_cast<const Reg&>(reg), 0x0F, preCode, code, (imm8 != NONE) ? 1 : 0);
+			opModM(op.getAddress(), reg.getReg(), 0x0F, preCode, code, (imm8 != NONE) ? 1 : 0);
 		} else {
-			opModR(static_cast<const Reg&>(reg), static_cast<const Reg&>(op), 0x0F, preCode, code);
+			opModR(reg.getReg(), op.getReg(), 0x0F, preCode, code);
 		}
 		if (imm8 != NONE) db(imm8);
 	}
 	void opMMX_IMM(const Mmx& mmx, int imm8, int code, int ext)
 	{
+		if (!isValidSSE(mmx)) XBYAK_THROW(ERR_NOT_SUPPORTED)
 		if (mmx.isXMM()) db(0x66);
 		opModR(Reg32(ext), mmx, 0x0F, code);
 		db(imm8);
@@ -1660,20 +2047,22 @@ private:
 	}
 	void opMovXMM(const Operand& op1, const Operand& op2, int code, int pref)
 	{
+		if (!isValidSSE(op1) || !isValidSSE(op2)) XBYAK_THROW(ERR_NOT_SUPPORTED)
 		if (pref != NONE) db(pref);
 		if (op1.isXMM() && op2.isMEM()) {
-			opModM(static_cast<const Address&>(op2), static_cast<const Reg&>(op1), 0x0F, code);
+			opModM(op2.getAddress(), op1.getReg(), 0x0F, code);
 		} else if (op1.isMEM() && op2.isXMM()) {
-			opModM(static_cast<const Address&>(op1), static_cast<const Reg&>(op2), 0x0F, code | 1);
+			opModM(op1.getAddress(), op2.getReg(), 0x0F, code | 1);
 		} else {
-			throw Error(ERR_BAD_COMBINATION);
+			XBYAK_THROW(ERR_BAD_COMBINATION)
 		}
 	}
 	void opExt(const Operand& op, const Mmx& mmx, int code, int imm, bool hasMMX2 = false)
 	{
+		if (!isValidSSE(op) || !isValidSSE(mmx)) XBYAK_THROW(ERR_NOT_SUPPORTED)
 		if (hasMMX2 && op.isREG(i32e)) { /* pextrw is special */
 			if (mmx.isXMM()) db(0x66);
-			opModR(static_cast<const Reg&>(op), mmx, 0x0F, 0xC5); db(imm);
+			opModR(op.getReg(), mmx, 0x0F, 0xC5); db(imm);
 		} else {
 			opGen(mmx, op, code, 0x66, isXMM_REG32orMEM, imm, 0x3A);
 		}
@@ -1683,11 +2072,11 @@ private:
 		int opBit = op.getBit();
 		if (disableRex && opBit == 64) opBit = 32;
 		if (op.isREG(bit)) {
-			opModR(Reg(ext, Operand::REG, opBit), static_cast<const Reg&>(op).changeBit(opBit), code0, code1, code2);
+			opModR(Reg(ext, Operand::REG, opBit), op.getReg().changeBit(opBit), code0, code1, code2);
 		} else if (op.isMEM()) {
-			opModM(static_cast<const Address&>(op), Reg(ext, Operand::REG, opBit), code0, code1, code2, immSize);
+			opModM(op.getAddress(), Reg(ext, Operand::REG, opBit), code0, code1, code2, immSize);
 		} else {
-			throw Error(ERR_BAD_COMBINATION);
+			XBYAK_THROW(ERR_BAD_COMBINATION)
 		}
 	}
 	void opShift(const Operand& op, int imm, int ext)
@@ -1698,22 +2087,22 @@ private:
 	}
 	void opShift(const Operand& op, const Reg8& _cl, int ext)
 	{
-		if (_cl.getIdx() != Operand::CL) throw Error(ERR_BAD_COMBINATION);
+		if (_cl.getIdx() != Operand::CL) XBYAK_THROW(ERR_BAD_COMBINATION)
 		opR_ModM(op, 0, ext, 0xD2);
 	}
 	void opModRM(const Operand& op1, const Operand& op2, bool condR, bool condM, int code0, int code1 = NONE, int code2 = NONE, int immSize = 0)
 	{
 		if (condR) {
-			opModR(static_cast<const Reg&>(op1), static_cast<const Reg&>(op2), code0, code1, code2);
+			opModR(op1.getReg(), op2.getReg(), code0, code1, code2);
 		} else if (condM) {
-			opModM(static_cast<const Address&>(op2), static_cast<const Reg&>(op1), code0, code1, code2, immSize);
+			opModM(op2.getAddress(), op1.getReg(), code0, code1, code2, immSize);
 		} else {
-			throw Error(ERR_BAD_COMBINATION);
+			XBYAK_THROW(ERR_BAD_COMBINATION)
 		}
 	}
-	void opShxd(const Operand& op, const Reg& reg, uint8 imm, int code, const Reg8 *_cl = 0)
+	void opShxd(const Operand& op, const Reg& reg, uint8_t imm, int code, const Reg8 *_cl = 0)
 	{
-		if (_cl && _cl->getIdx() != Operand::CL) throw Error(ERR_BAD_COMBINATION);
+		if (_cl && _cl->getIdx() != Operand::CL) XBYAK_THROW(ERR_BAD_COMBINATION)
 		opModRM(reg, op, (op.isREG(16 | i32e) && op.getBit() == reg.getBit()), op.isMEM() && (reg.isREG(16 | i32e)), 0x0F, code | (_cl ? 1 : 0), NONE, _cl ? 0 : 1);
 		if (!_cl) db(imm);
 	}
@@ -1721,18 +2110,18 @@ private:
 	void opRM_RM(const Operand& op1, const Operand& op2, int code)
 	{
 		if (op1.isREG() && op2.isMEM()) {
-			opModM(static_cast<const Address&>(op2), static_cast<const Reg&>(op1), code | 2);
+			opModM(op2.getAddress(), op1.getReg(), code | 2);
 		} else {
 			opModRM(op2, op1, op1.isREG() && op1.getKind() == op2.getKind(), op1.isMEM() && op2.isREG(), code);
 		}
 	}
 	// (REG|MEM, IMM)
-	void opRM_I(const Operand& op, uint32 imm, int code, int ext)
+	void opRM_I(const Operand& op, uint32_t imm, int code, int ext)
 	{
 		verifyMemHasSize(op);
-		uint32 immBit = inner::IsInDisp8(imm) ? 8 : isInDisp16(imm) ? 16 : 32;
+		uint32_t immBit = inner::IsInDisp8(imm) ? 8 : isInDisp16(imm) ? 16 : 32;
 		if (op.isBit(8)) immBit = 8;
-		if (op.getBit() < immBit) throw Error(ERR_IMM_IS_TOO_BIG);
+		if (op.getBit() < immBit) XBYAK_THROW(ERR_IMM_IS_TOO_BIG)
 		if (op.isBit(32|64) && immBit == 16) immBit = 32; /* don't use MEM16 if 32/64bit mode */
 		if (op.isREG() && op.getIdx() == 0 && (op.getBit() == immBit || (op.isBit(64) && immBit == 32))) { // rax, eax, ax, al
 			rex(op);
@@ -1754,36 +2143,41 @@ private:
 #endif
 		code = 0xFE;
 		if (op.isREG()) {
-			opModR(Reg(ext, Operand::REG, op.getBit()), static_cast<const Reg&>(op), code);
+			opModR(Reg(ext, Operand::REG, op.getBit()), op.getReg(), code);
 		} else {
-			opModM(static_cast<const Address&>(op), Reg(ext, Operand::REG, op.getBit()), code);
+			opModM(op.getAddress(), Reg(ext, Operand::REG, op.getBit()), code);
 		}
 	}
 	void opPushPop(const Operand& op, int code, int ext, int alt)
 	{
-		if (op.isREG()) {
-			if (op.isBit(16)) db(0x66);
-			if (static_cast<const Reg&>(op).getIdx() >= 8) db(0x41);
-			db(alt | (op.getIdx() & 7));
-		} else if (op.isMEM()) {
-			opModM(static_cast<const Address&>(op), Reg(ext, Operand::REG, op.getBit()), code);
-		} else {
-			throw Error(ERR_BAD_COMBINATION);
+		int bit = op.getBit();
+		if (bit == 16 || bit == BIT) {
+			if (bit == 16) db(0x66);
+			if (op.isREG()) {
+				if (op.getReg().getIdx() >= 8) db(0x41);
+				db(alt | (op.getIdx() & 7));
+				return;
+			}
+			if (op.isMEM()) {
+				opModM(op.getAddress(), Reg(ext, Operand::REG, 32), code);
+				return;
+			}
 		}
+		XBYAK_THROW(ERR_BAD_COMBINATION)
 	}
 	void verifyMemHasSize(const Operand& op) const
 	{
-		if (op.isMEM() && op.getBit() == 0) throw Error(ERR_MEM_SIZE_IS_NOT_SPECIFIED);
+		if (op.isMEM() && op.getBit() == 0) XBYAK_THROW(ERR_MEM_SIZE_IS_NOT_SPECIFIED)
 	}
 	/*
 		mov(r, imm) = db(imm, mov_imm(r, imm))
 	*/
-	int mov_imm(const Reg& reg, size_t imm)
+	int mov_imm(const Reg& reg, uint64_t imm)
 	{
 		int bit = reg.getBit();
 		const int idx = reg.getIdx();
 		int code = 0xB0 | ((bit == 8 ? 0 : 1) << 3);
-		if (bit == 64 && (imm & ~size_t(0xffffffffu)) == 0) {
+		if (bit == 64 && (imm & ~uint64_t(0xffffffffu)) == 0) {
 			rex(Reg32(idx));
 			bit = 32;
 		} else {
@@ -1807,32 +2201,29 @@ private:
 			if (relative) {
 				db(inner::VerifyInInt32(offset + disp - size_ - jmpSize), jmpSize);
 			} else if (isAutoGrow()) {
-				db(uint64(0), jmpSize);
+				db(uint64_t(0), jmpSize);
 				save(size_ - jmpSize, offset, jmpSize, inner::LaddTop);
 			} else {
 				db(size_t(top_) + offset, jmpSize);
 			}
 			return;
 		}
-		db(uint64(0), jmpSize);
+		db(uint64_t(0), jmpSize);
 		JmpLabel jmp(size_, jmpSize, (relative ? inner::LasIs : isAutoGrow() ? inner::LaddTop : inner::Labs), disp);
 		labelMgr_.addUndefinedLabel(label, jmp);
 	}
-	void opMovxx(const Reg& reg, const Operand& op, uint8 code)
+	void opMovxx(const Reg& reg, const Operand& op, uint8_t code)
 	{
-		if (op.isBit(32)) throw Error(ERR_BAD_COMBINATION);
+		if (op.isBit(32)) XBYAK_THROW(ERR_BAD_COMBINATION)
 		int w = op.isBit(16);
-#ifdef XBYAK64
-		if (op.isHigh8bit()) throw Error(ERR_BAD_COMBINATION);
-#endif
 		bool cond = reg.isREG() && (reg.getBit() > op.getBit());
 		opModRM(reg, op, cond && op.isREG(), cond && op.isMEM(), 0x0F, code | w);
 	}
-	void opFpuMem(const Address& addr, uint8 m16, uint8 m32, uint8 m64, uint8 ext, uint8 m64ext)
+	void opFpuMem(const Address& addr, uint8_t m16, uint8_t m32, uint8_t m64, uint8_t ext, uint8_t m64ext)
 	{
-		if (addr.is64bitDisp()) throw Error(ERR_CANT_USE_64BIT_DISP);
-		uint8 code = addr.isBit(16) ? m16 : addr.isBit(32) ? m32 : addr.isBit(64) ? m64 : 0;
-		if (!code) throw Error(ERR_BAD_MEM_SIZE);
+		if (addr.is64bitDisp()) XBYAK_THROW(ERR_CANT_USE_64BIT_DISP)
+		uint8_t code = addr.isBit(16) ? m16 : addr.isBit(32) ? m32 : addr.isBit(64) ? m64 : 0;
+		if (!code) XBYAK_THROW(ERR_BAD_MEM_SIZE)
 		if (m64ext && addr.isBit(64)) ext = m64ext;
 
 		rex(addr, st0);
@@ -1841,41 +2232,43 @@ private:
 	}
 	// use code1 if reg1 == st0
 	// use code2 if reg1 != st0 && reg2 == st0
-	void opFpuFpu(const Fpu& reg1, const Fpu& reg2, uint32 code1, uint32 code2)
+	void opFpuFpu(const Fpu& reg1, const Fpu& reg2, uint32_t code1, uint32_t code2)
 	{
-		uint32 code = reg1.getIdx() == 0 ? code1 : reg2.getIdx() == 0 ? code2 : 0;
-		if (!code) throw Error(ERR_BAD_ST_COMBINATION);
-		db(uint8(code >> 8));
-		db(uint8(code | (reg1.getIdx() | reg2.getIdx())));
+		uint32_t code = reg1.getIdx() == 0 ? code1 : reg2.getIdx() == 0 ? code2 : 0;
+		if (!code) XBYAK_THROW(ERR_BAD_ST_COMBINATION)
+		db(uint8_t(code >> 8));
+		db(uint8_t(code | (reg1.getIdx() | reg2.getIdx())));
 	}
-	void opFpu(const Fpu& reg, uint8 code1, uint8 code2)
+	void opFpu(const Fpu& reg, uint8_t code1, uint8_t code2)
 	{
 		db(code1); db(code2 | reg.getIdx());
 	}
 	void opVex(const Reg& r, const Operand *p1, const Operand& op2, int type, int code, int imm8 = NONE)
 	{
 		if (op2.isMEM()) {
-			const Address& addr = static_cast<const Address&>(op2);
-			const Reg& base = addr.getRegExp().getBase();
+			const Address& addr = op2.getAddress();
+			const RegExp& regExp = addr.getRegExp();
+			const Reg& base = regExp.getBase();
+			const Reg& index = regExp.getIndex();
 			if (BIT == 64 && addr.is32bit()) db(0x67);
 			int disp8N = 0;
-			bool x = addr.getRegExp().getIndex().isExtIdx();
-			if ((type & T_MUST_EVEX) || r.hasEvex() || (p1 && p1->hasEvex()) || addr.isBroadcast() || addr.getOpmaskIdx()) {
+			bool x = index.isExtIdx();
+			if ((type & (T_MUST_EVEX|T_MEM_EVEX)) || r.hasEvex() || (p1 && p1->hasEvex()) || addr.isBroadcast() || addr.getOpmaskIdx()) {
 				int aaa = addr.getOpmaskIdx();
-				if (aaa && !(type & T_M_K)) throw Error(ERR_INVALID_OPMASK_WITH_MEMORY);
+				if (aaa && !(type & T_M_K)) XBYAK_THROW(ERR_INVALID_OPMASK_WITH_MEMORY)
 				bool b = false;
 				if (addr.isBroadcast()) {
-					if (!(type & (T_B32 | T_B64))) throw Error(ERR_INVALID_BROADCAST);
+					if (!(type & (T_B32 | T_B64))) XBYAK_THROW(ERR_INVALID_BROADCAST)
 					b = true;
 				}
-				int VL = addr.getRegExp().isVsib() ? addr.getRegExp().getIndex().getBit() : 0;
-				disp8N = evex(r, base, p1, type, code, x, b, aaa, VL);
+				int VL = regExp.isVsib() ? index.getBit() : 0;
+				disp8N = evex(r, base, p1, type, code, x, b, aaa, VL, index.isExtIdx2());
 			} else {
 				vex(r, base, p1, type, code, x);
 			}
-			opAddr(addr, r.getIdx(), (imm8 != NONE) ? 1 : 0, disp8N);
+			opAddr(addr, r.getIdx(), (imm8 != NONE) ? 1 : 0, disp8N, (type & T_VSIB) != 0);
 		} else {
-			const Reg& base = static_cast<const Reg&>(op2);
+			const Reg& base = op2.getReg();
 			if ((type & T_MUST_EVEX) || r.hasEvex() || (p1 && p1->hasEvex()) || base.hasEvex()) {
 				evex(r, base, p1, type, code);
 			} else {
@@ -1887,13 +2280,13 @@ private:
 	}
 	// (r, r, r/m) if isR_R_RM
 	// (r, r/m, r)
-	void opGpr(const Reg32e& r, const Operand& op1, const Operand& op2, int type, uint8 code, bool isR_R_RM, int imm8 = NONE)
+	void opGpr(const Reg32e& r, const Operand& op1, const Operand& op2, int type, uint8_t code, bool isR_R_RM, int imm8 = NONE)
 	{
 		const Operand *p1 = &op1;
 		const Operand *p2 = &op2;
 		if (!isR_R_RM) std::swap(p1, p2);
 		const unsigned int bit = r.getBit();
-		if (p1->getBit() != bit || (p2->isREG() && p2->getBit() != bit)) throw Error(ERR_BAD_COMBINATION);
+		if (p1->getBit() != bit || (p2->isREG() && p2->getBit() != bit)) XBYAK_THROW(ERR_BAD_COMBINATION)
 		type |= (bit == 64) ? T_W1 : T_W0;
 		opVex(r, p1, *p2, type, code, imm8);
 	}
@@ -1906,37 +2299,53 @@ private:
 			op = &op1;
 		}
 		// (x1, x2, op)
-		if (!((x1.isXMM() && x2->isXMM()) || ((type & T_YMM) && ((x1.isYMM() && x2->isYMM()) || (x1.isZMM() && x2->isZMM()))))) throw Error(ERR_BAD_COMBINATION);
+		if (!((x1.isXMM() && x2->isXMM()) || ((type & T_YMM) && ((x1.isYMM() && x2->isYMM()) || (x1.isZMM() && x2->isZMM()))))) XBYAK_THROW(ERR_BAD_COMBINATION)
 		opVex(x1, x2, *op, type, code0, imm8);
 	}
 	void opAVX_K_X_XM(const Opmask& k, const Xmm& x2, const Operand& op3, int type, int code0, int imm8 = NONE)
 	{
-		if (!op3.isMEM() && (x2.getKind() != op3.getKind())) throw Error(ERR_BAD_COMBINATION);
+		if (!op3.isMEM() && (x2.getKind() != op3.getKind())) XBYAK_THROW(ERR_BAD_COMBINATION)
 		opVex(k, &x2, op3, type, code0, imm8);
 	}
 	// (x, x/m), (y, x/m256), (z, y/m)
 	void checkCvt1(const Operand& x, const Operand& op) const
 	{
-		if (!op.isMEM() && !(x.is(Operand::XMM | Operand::YMM) && op.isXMM()) && !(x.isZMM() && op.isYMM())) throw Error(ERR_BAD_COMBINATION);
+		if (!op.isMEM() && !(x.is(Operand::XMM | Operand::YMM) && op.isXMM()) && !(x.isZMM() && op.isYMM())) XBYAK_THROW(ERR_BAD_COMBINATION)
 	}
 	// (x, x/m), (x, y/m256), (y, z/m)
 	void checkCvt2(const Xmm& x, const Operand& op) const
 	{
-		if (!(x.isXMM() && op.is(Operand::XMM | Operand::YMM | Operand::MEM)) && !(x.isYMM() && op.is(Operand::ZMM | Operand::MEM))) throw Error(ERR_BAD_COMBINATION);
+		if (!(x.isXMM() && op.is(Operand::XMM | Operand::YMM | Operand::MEM)) && !(x.isYMM() && op.is(Operand::ZMM | Operand::MEM))) XBYAK_THROW(ERR_BAD_COMBINATION)
 	}
-	void opCvt2(const Xmm& x, const Operand& op, int type, int code)
+	void opCvt(const Xmm& x, const Operand& op, int type, int code)
 	{
-		checkCvt2(x, op);
 		Operand::Kind kind = x.isXMM() ? (op.isBit(256) ? Operand::YMM : Operand::XMM) : Operand::ZMM;
 		opVex(x.copyAndSetKind(kind), &xm0, op, type, code);
 	}
-	void opCvt3(const Xmm& x1, const Xmm& x2, const Operand& op, int type, int type64, int type32, uint8 code)
+	void opCvt2(const Xmm& x, const Operand& op, int type, int code)
+	{
+		checkCvt2(x, op);
+		opCvt(x, op, type, code);
+	}
+	void opCvt3(const Xmm& x1, const Xmm& x2, const Operand& op, int type, int type64, int type32, uint8_t code)
 	{
-		if (!(x1.isXMM() && x2.isXMM() && (op.isREG(i32e) || op.isMEM()))) throw Error(ERR_BAD_SIZE_OF_REGISTER);
+		if (!(x1.isXMM() && x2.isXMM() && (op.isREG(i32e) || op.isMEM()))) XBYAK_THROW(ERR_BAD_SIZE_OF_REGISTER)
 		Xmm x(op.getIdx());
 		const Operand *p = op.isREG() ? &x : &op;
 		opVex(x1, &x2, *p, type | (op.isBit(64) ? type64 : type32), code);
 	}
+	// (x, x/y/xword/yword), (y, z/m)
+	void checkCvt4(const Xmm& x, const Operand& op) const
+	{
+		if (!(x.isXMM() && op.is(Operand::XMM | Operand::YMM | Operand::MEM) && op.isBit(128|256)) && !(x.isYMM() && op.is(Operand::ZMM | Operand::MEM))) XBYAK_THROW(ERR_BAD_COMBINATION)
+	}
+	// (x, x/y/z/xword/yword/zword)
+	void opCvt5(const Xmm& x, const Operand& op, int type, int code)
+	{
+		if (!(x.isXMM() && op.isBit(128|256|512))) XBYAK_THROW(ERR_BAD_COMBINATION)
+		Operand::Kind kind = op.isBit(128) ? Operand::XMM : op.isBit(256) ? Operand::YMM : Operand::ZMM;
+		opVex(x.copyAndSetKind(kind), &xm0, op, type, code);
+	}
 	const Xmm& cvtIdx0(const Operand& x) const
 	{
 		return x.isZMM() ? zm0 : x.isYMM() ? ym0 : xm0;
@@ -1947,21 +2356,22 @@ private:
 		opAVX_X_X_XM(x, cvtIdx0(x), op, type, code, imm8);
 	}
 	// QQQ:need to refactor
-	void opSp1(const Reg& reg, const Operand& op, uint8 pref, uint8 code0, uint8 code1)
+	void opSp1(const Reg& reg, const Operand& op, uint8_t pref, uint8_t code0, uint8_t code1)
 	{
-		if (reg.isBit(8)) throw Error(ERR_BAD_SIZE_OF_REGISTER);
+		if (reg.isBit(8)) XBYAK_THROW(ERR_BAD_SIZE_OF_REGISTER)
 		bool is16bit = reg.isREG(16) && (op.isREG(16) || op.isMEM());
-		if (!is16bit && !(reg.isREG(i32e) && (op.isREG(reg.getBit()) || op.isMEM()))) throw Error(ERR_BAD_COMBINATION);
+		if (!is16bit && !(reg.isREG(i32e) && (op.isREG(reg.getBit()) || op.isMEM()))) XBYAK_THROW(ERR_BAD_COMBINATION)
 		if (is16bit) db(0x66);
 		db(pref); opModRM(reg.changeBit(i32e == 32 ? 32 : reg.getBit()), op, op.isREG(), true, code0, code1);
 	}
-	void opGather(const Xmm& x1, const Address& addr, const Xmm& x2, int type, uint8 code, int mode)
+	void opGather(const Xmm& x1, const Address& addr, const Xmm& x2, int type, uint8_t code, int mode)
 	{
-		if (!addr.getRegExp().isVsib(128 | 256)) throw Error(ERR_BAD_VSIB_ADDRESSING);
+		const RegExp& regExp = addr.getRegExp();
+		if (!regExp.isVsib(128 | 256)) XBYAK_THROW(ERR_BAD_VSIB_ADDRESSING)
 		const int y_vx_y = 0;
 		const int y_vy_y = 1;
 //		const int x_vy_x = 2;
-		const bool isAddrYMM = addr.getRegExp().getIndex().getBit() == 256;
+		const bool isAddrYMM = regExp.getIndex().getBit() == 256;
 		if (!x1.isXMM() || isAddrYMM || !x2.isXMM()) {
 			bool isOK = false;
 			if (mode == y_vx_y) {
@@ -1971,10 +2381,13 @@ private:
 			} else { // x_vy_x
 				isOK = !x1.isYMM() && isAddrYMM && !x2.isYMM();
 			}
-			if (!isOK) throw Error(ERR_BAD_VSIB_ADDRESSING);
+			if (!isOK) XBYAK_THROW(ERR_BAD_VSIB_ADDRESSING)
 		}
-		addr.permitVsib();
-		opAVX_X_X_XM(isAddrYMM ? Ymm(x1.getIdx()) : x1, isAddrYMM ? Ymm(x2.getIdx()) : x2, addr, type | T_YMM, code);
+		int i1 = x1.getIdx();
+		int i2 = regExp.getIndex().getIdx();
+		int i3 = x2.getIdx();
+		if (i1 == i2 || i1 == i3 || i2 == i3) XBYAK_THROW(ERR_SAME_REGS_ARE_INVALID);
+		opAVX_X_X_XM(isAddrYMM ? Ymm(i1) : x1, isAddrYMM ? Ymm(i3) : x2, addr, type, code);
 	}
 	enum {
 		xx_yy_zz = 0,
@@ -1992,35 +2405,85 @@ private:
 		case xx_xy_yz: if ((x1.isXMM() && x2.isYMM()) || (x1.isYMM() && x2.isZMM())) return;
 			break;
 		}
-		throw Error(ERR_BAD_VSIB_ADDRESSING);
+		XBYAK_THROW(ERR_BAD_VSIB_ADDRESSING)
 	}
-	void opGather2(const Xmm& x, const Address& addr, int type, uint8 code, int mode)
+	void opGather2(const Xmm& x, const Address& addr, int type, uint8_t code, int mode)
 	{
-		if (x.hasZero()) throw Error(ERR_INVALID_ZERO);
-		checkGather2(x, addr.getRegExp().getIndex(), mode);
-		addr.permitVsib();
+		if (x.hasZero()) XBYAK_THROW(ERR_INVALID_ZERO)
+		const RegExp& regExp = addr.getRegExp();
+		checkGather2(x, regExp.getIndex(), mode);
+		int maskIdx = x.getOpmaskIdx();
+		if ((type & T_M_K) && addr.getOpmaskIdx()) maskIdx = addr.getOpmaskIdx();
+		if (maskIdx == 0) XBYAK_THROW(ERR_K0_IS_INVALID);
+		if (!(type & T_M_K) && x.getIdx() == regExp.getIndex().getIdx()) XBYAK_THROW(ERR_SAME_REGS_ARE_INVALID);
 		opVex(x, 0, addr, type, code);
 	}
 	/*
 		xx_xy_yz ; mode = true
 		xx_xy_xz ; mode = false
 	*/
-	void opVmov(const Operand& op, const Xmm& x, int type, uint8 code, bool mode)
+	void opVmov(const Operand& op, const Xmm& x, int type, uint8_t code, bool mode)
 	{
 		if (mode) {
-			if (!op.isMEM() && !((op.isXMM() && x.isXMM()) || (op.isXMM() && x.isYMM()) || (op.isYMM() && x.isZMM())))  throw Error(ERR_BAD_COMBINATION);
+			if (!op.isMEM() && !((op.isXMM() && x.isXMM()) || (op.isXMM() && x.isYMM()) || (op.isYMM() && x.isZMM()))) XBYAK_THROW(ERR_BAD_COMBINATION)
 		} else {
-			if (!op.isMEM() && !op.isXMM()) throw Error(ERR_BAD_COMBINATION);
+			if (!op.isMEM() && !op.isXMM()) XBYAK_THROW(ERR_BAD_COMBINATION)
 		}
 		opVex(x, 0, op, type, code);
 	}
-	void opGatherFetch(const Address& addr, const Xmm& x, int type, uint8 code, Operand::Kind kind)
+	void opGatherFetch(const Address& addr, const Xmm& x, int type, uint8_t code, Operand::Kind kind)
 	{
-		if (addr.hasZero()) throw Error(ERR_INVALID_ZERO);
-		if (addr.getRegExp().getIndex().getKind() != kind) throw Error(ERR_BAD_VSIB_ADDRESSING);
-		addr.permitVsib();
+		if (addr.hasZero()) XBYAK_THROW(ERR_INVALID_ZERO)
+		if (addr.getRegExp().getIndex().getKind() != kind) XBYAK_THROW(ERR_BAD_VSIB_ADDRESSING)
 		opVex(x, 0, addr, type, code);
 	}
+	void opEncoding(const Xmm& x1, const Xmm& x2, const Operand& op, int type, int code0, PreferredEncoding encoding)
+	{
+		opAVX_X_X_XM(x1, x2, op, type | orEvexIf(encoding), code0);
+	}
+	int orEvexIf(PreferredEncoding encoding) {
+		if (encoding == DefaultEncoding) {
+			encoding = defaultEncoding_;
+		}
+		if (encoding == EvexEncoding) {
+#ifdef XBYAK_DISABLE_AVX512
+			XBYAK_THROW(ERR_EVEX_IS_INVALID)
+#endif
+			return T_MUST_EVEX;
+		}
+		return 0;
+	}
+	void opInOut(const Reg& a, const Reg& d, uint8_t code)
+	{
+		if (a.getIdx() == Operand::AL && d.getIdx() == Operand::DX && d.getBit() == 16) {
+			switch (a.getBit()) {
+			case 8: db(code); return;
+			case 16: db(0x66); db(code + 1); return;
+			case 32: db(code + 1); return;
+			}
+		}
+		XBYAK_THROW(ERR_BAD_COMBINATION)
+	}
+	void opInOut(const Reg& a, uint8_t code, uint8_t v)
+	{
+		if (a.getIdx() == Operand::AL) {
+			switch (a.getBit()) {
+			case 8: db(code); db(v); return;
+			case 16: db(0x66); db(code + 1); db(v); return;
+			case 32: db(code + 1); db(v); return;
+			}
+		}
+		XBYAK_THROW(ERR_BAD_COMBINATION)
+	}
+#ifdef XBYAK64
+	void opAMX(const Tmm& t1, const Address& addr, int type, int code0)
+	{
+		// require both base and index
+		const RegExp exp = addr.getRegExp(false);
+		if (exp.getBase().getBit() == 0 || exp.getIndex().getBit() == 0) XBYAK_THROW(ERR_NOT_SUPPORTED)
+		opVex(t1, &tmm0, addr, type, code0);
+	}
+#endif
 public:
 	unsigned int getVersion() const { return VERSION; }
 	using CodeArray::db;
@@ -2030,7 +2493,7 @@ public:
 	const Zmm zmm0, zmm1, zmm2, zmm3, zmm4, zmm5, zmm6, zmm7;
 	const Xmm &xm0, &xm1, &xm2, &xm3, &xm4, &xm5, &xm6, &xm7;
 	const Ymm &ym0, &ym1, &ym2, &ym3, &ym4, &ym5, &ym6, &ym7;
-	const Ymm &zm0, &zm1, &zm2, &zm3, &zm4, &zm5, &zm6, &zm7;
+	const Zmm &zm0, &zm1, &zm2, &zm3, &zm4, &zm5, &zm6, &zm7;
 	const Reg32 eax, ecx, edx, ebx, esp, ebp, esi, edi;
 	const Reg16 ax, cx, dx, bx, sp, bp, si, di;
 	const Reg8 al, cl, dl, bl, ah, ch, dh, bh;
@@ -2038,6 +2501,7 @@ public:
 	const AddressFrame ptr_b, xword_b, yword_b, zword_b; // broadcast such as {1to2}, {1to4}, {1to8}, {1to16}, {b}
 	const Fpu st0, st1, st2, st3, st4, st5, st6, st7;
 	const Opmask k0, k1, k2, k3, k4, k5, k6, k7;
+	const BoundsReg bnd0, bnd1, bnd2, bnd3;
 	const EvexModifierRounding T_sae, T_rn_sae, T_rd_sae, T_ru_sae, T_rz_sae; // {sae}, {rn-sae}, {rd-sae}, {ru-sae}, {rz-sae}
 	const EvexModifierZero T_z; // {z}
 #ifdef XBYAK64
@@ -2055,6 +2519,7 @@ public:
 	const Zmm zmm8, zmm9, zmm10, zmm11, zmm12, zmm13, zmm14, zmm15;
 	const Zmm zmm16, zmm17, zmm18, zmm19, zmm20, zmm21, zmm22, zmm23;
 	const Zmm zmm24, zmm25, zmm26, zmm27, zmm28, zmm29, zmm30, zmm31;
+	const Tmm tmm0, tmm1, tmm2, tmm3, tmm4, tmm5, tmm6, tmm7;
 	const Xmm &xm8, &xm9, &xm10, &xm11, &xm12, &xm13, &xm14, &xm15; // for my convenience
 	const Xmm &xm16, &xm17, &xm18, &xm19, &xm20, &xm21, &xm22, &xm23;
 	const Xmm &xm24, &xm25, &xm26, &xm27, &xm28, &xm29, &xm30, &xm31;
@@ -2069,8 +2534,13 @@ public:
 #ifndef XBYAK_DISABLE_SEGMENT
 	const Segment es, cs, ss, ds, fs, gs;
 #endif
+private:
+	bool isDefaultJmpNEAR_;
+	PreferredEncoding defaultEncoding_;
+public:
 	void L(const std::string& label) { labelMgr_.defineSlabel(label); }
-	void L(const Label& label) { labelMgr_.defineClabel(label); }
+	void L(Label& label) { labelMgr_.defineClabel(label); }
+	Label L() { Label label; L(label); return label; }
 	void inLocalLabel() { labelMgr_.enterLocal(); }
 	void outLocalLabel() { labelMgr_.leaveLocal(); }
 	/*
@@ -2087,13 +2557,15 @@ public:
 	void putL(std::string label) { putL_inner(label); }
 	void putL(const Label& label) { putL_inner(label); }
 
-	void jmp(const Operand& op) { opR_ModM(op, BIT, 4, 0xFF, NONE, NONE, true); }
+	// set default type of `jmp` of undefined label to T_NEAR
+	void setDefaultJmpNEAR(bool isNear) { isDefaultJmpNEAR_ = isNear; }
+	void jmp(const Operand& op, LabelType type = T_AUTO) { opJmpOp(op, type, 4); }
 	void jmp(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0xEB, 0xE9, 0); }
 	void jmp(const char *label, LabelType type = T_AUTO) { jmp(std::string(label), type); }
 	void jmp(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0xEB, 0xE9, 0); }
 	void jmp(const void *addr, LabelType type = T_AUTO) { opJmpAbs(addr, type, 0xEB, 0xE9); }
 
-	void call(const Operand& op) { opR_ModM(op, 16 | i32e, 2, 0xFF, NONE, NONE, true); }
+	void call(const Operand& op, LabelType type = T_AUTO) { opJmpOp(op, type, 2); }
 	// call(string label), not const std::string&
 	void call(std::string label) { opJmp(label, T_NEAR, 0, 0xE8, 0); }
 	void call(const char *label) { call(std::string(label)); }
@@ -2101,7 +2573,7 @@ public:
 	// call(function pointer)
 #ifdef XBYAK_VARIADIC_TEMPLATE
 	template<class Ret, class... Params>
-	void call(Ret(*func)(Params...)) { call(CastTo<const void*>(func)); }
+	void call(Ret(*func)(Params...)) { call(reinterpret_cast<const void*>(func)); }
 #endif
 	void call(const void *addr) { opJmpAbs(addr, T_NEAR, 0, 0xE8); }
 
@@ -2109,7 +2581,7 @@ public:
 	{
 		opModRM(reg, op, op.isREG() && (op.getKind() == reg.getKind()), op.isMEM(), 0x84);
 	}
-	void test(const Operand& op, uint32 imm)
+	void test(const Operand& op, uint32_t imm)
 	{
 		verifyMemHasSize(op);
         int immSize = (std::min)(op.getBit() / 8, 4U);
@@ -2134,18 +2606,18 @@ public:
 	}
 	void push(const Operand& op) { opPushPop(op, 0xFF, 6, 0x50); }
 	void pop(const Operand& op) { opPushPop(op, 0x8F, 0, 0x58); }
-	void push(const AddressFrame& af, uint32 imm)
+	void push(const AddressFrame& af, uint32_t imm)
 	{
-		if (af.bit_ == 8 && inner::IsInDisp8(imm)) {
+		if (af.bit_ == 8) {
 			db(0x6A); db(imm);
-		} else if (af.bit_ == 16 && isInDisp16(imm)) {
+		} else if (af.bit_ == 16) {
 			db(0x66); db(0x68); dw(imm);
 		} else {
 			db(0x68); dd(imm);
 		}
 	}
 	/* use "push(word, 4)" if you want "push word 4" */
-	void push(uint32 imm)
+	void push(uint32_t imm)
 	{
 		if (inner::IsInDisp8(imm)) {
 			push(byte, imm);
@@ -2157,15 +2629,15 @@ public:
 	{
 		const Reg *reg = 0;
 		const Address *addr = 0;
-		uint8 code = 0;
+		uint8_t code = 0;
 		if (reg1.isREG() && reg1.getIdx() == 0 && reg2.isMEM()) { // mov eax|ax|al, [disp]
-			reg = &static_cast<const Reg&>(reg1);
-			addr= &static_cast<const Address&>(reg2);
+			reg = &reg1.getReg();
+			addr= &reg2.getAddress();
 			code = 0xA0;
 		} else
 		if (reg1.isMEM() && reg2.isREG() && reg2.getIdx() == 0) { // mov [disp], eax|ax|al
-			reg = &static_cast<const Reg&>(reg2);
-			addr= &static_cast<const Address&>(reg1);
+			reg = &reg2.getReg();
+			addr= &reg1.getAddress();
 			code = 0xA2;
 		}
 #ifdef XBYAK64
@@ -2175,50 +2647,47 @@ public:
 				db(reg1.isREG(8) ? 0xA0 : reg1.isREG() ? 0xA1 : reg2.isREG(8) ? 0xA2 : 0xA3);
 				db(addr->getDisp(), 8);
 			} else {
-				throw Error(ERR_BAD_COMBINATION);
+				XBYAK_THROW(ERR_BAD_COMBINATION)
 			}
 		} else
 #else
 		if (code && addr->isOnlyDisp()) {
 			rex(*reg, *addr);
 			db(code | (reg->isBit(8) ? 0 : 1));
-			dd(static_cast<uint32>(addr->getDisp()));
+			dd(static_cast<uint32_t>(addr->getDisp()));
 		} else
 #endif
 		{
 			opRM_RM(reg1, reg2, 0x88);
 		}
 	}
-	void mov(const Operand& op, size_t imm)
+	void mov(const Operand& op, uint64_t imm)
 	{
 		if (op.isREG()) {
-			const int size = mov_imm(static_cast<const Reg&>(op), imm);
+			const int size = mov_imm(op.getReg(), imm);
 			db(imm, size);
 		} else if (op.isMEM()) {
 			verifyMemHasSize(op);
 			int immSize = op.getBit() / 8;
 			if (immSize <= 4) {
-				sint64 s = sint64(imm) >> (immSize * 8);
-				if (s != 0 && s != -1) throw Error(ERR_IMM_IS_TOO_BIG);
+				int64_t s = int64_t(imm) >> (immSize * 8);
+				if (s != 0 && s != -1) XBYAK_THROW(ERR_IMM_IS_TOO_BIG)
 			} else {
-				if (!inner::IsInInt32(imm)) throw Error(ERR_IMM_IS_TOO_BIG);
+				if (!inner::IsInInt32(imm)) XBYAK_THROW(ERR_IMM_IS_TOO_BIG)
 				immSize = 4;
 			}
-			opModM(static_cast<const Address&>(op), Reg(0, Operand::REG, op.getBit()), 0xC6, NONE, NONE, immSize);
-			db(static_cast<uint32>(imm), immSize);
+			opModM(op.getAddress(), Reg(0, Operand::REG, op.getBit()), 0xC6, NONE, NONE, immSize);
+			db(static_cast<uint32_t>(imm), immSize);
 		} else {
-			throw Error(ERR_BAD_COMBINATION);
-		}
-	}
-	void mov(const NativeReg& reg, const char *label) // can't use std::string
-	{
-		if (label == 0) {
-			mov(static_cast<const Operand&>(reg), 0); // call imm
-			return;
+			XBYAK_THROW(ERR_BAD_COMBINATION)
 		}
-		mov_imm(reg, dummyAddr);
-		putL(label);
 	}
+
+	// The template is used to avoid ambiguity when the 2nd argument is 0.
+	// When the 2nd argument is 0 the call goes to
+	// `void mov(const Operand& op, uint64_t imm)`.
+	template <typename T1, typename T2>
+	void mov(const T1&, const T2 *) { T1::unexpected; }
 	void mov(const NativeReg& reg, const Label& label)
 	{
 		mov_imm(reg, dummyAddr);
@@ -2230,7 +2699,7 @@ public:
 		if (p1->isMEM() || (p2->isREG(16 | i32e) && p2->getIdx() == 0)) {
 			p1 = &op2; p2 = &op1;
 		}
-		if (p1->isMEM()) throw Error(ERR_BAD_COMBINATION);
+		if (p1->isMEM()) XBYAK_THROW(ERR_BAD_COMBINATION)
 		if (p2->isREG() && (p1->isREG(16 | i32e) && p1->getIdx() == 0)
 #ifdef XBYAK64
 			&& (p2->getIdx() != 0 || !p1->isREG(32))
@@ -2260,7 +2729,7 @@ public:
 	{
 		switch (seg.getIdx()) {
 		case Segment::es: db(0x07); break;
-		case Segment::cs: throw Error(ERR_BAD_COMBINATION);
+		case Segment::cs: XBYAK_THROW(ERR_BAD_COMBINATION)
 		case Segment::ss: db(0x17); break;
 		case Segment::ds: db(0x1F); break;
 		case Segment::fs: db(0x0F); db(0xA1); break;
@@ -2288,7 +2757,7 @@ public:
 	}
 	void mov(const Segment& seg, const Operand& op)
 	{
-		opModRM(Reg8(seg.getIdx()), op.isREG(16|i32e) ? static_cast<const Operand&>(static_cast<const Reg&>(op).cvt32()) : op, op.isREG(16|i32e), op.isMEM(), 0x8E);
+		opModRM(Reg8(seg.getIdx()), op.isREG(16|i32e) ? static_cast<const Operand&>(op.getReg().cvt32()) : op, op.isREG(16|i32e), op.isMEM(), 0x8E);
 	}
 #endif
 
@@ -2312,7 +2781,8 @@ public:
 		, ptr_b(0, true), xword_b(128, true), yword_b(256, true), zword_b(512, true)
 		, st0(0), st1(1), st2(2), st3(3), st4(4), st5(5), st6(6), st7(7)
 		, k0(0), k1(1), k2(2), k3(3), k4(4), k5(5), k6(6), k7(7)
-		, T_sae(T_SAE), T_rn_sae(T_RN_SAE), T_rd_sae(T_RD_SAE), T_ru_sae(T_RU_SAE), T_rz_sae(T_RZ_SAE)
+		, bnd0(0), bnd1(1), bnd2(2), bnd3(3)
+		, T_sae(EvexModifierRounding::T_SAE), T_rn_sae(EvexModifierRounding::T_RN_SAE), T_rd_sae(EvexModifierRounding::T_RD_SAE), T_ru_sae(EvexModifierRounding::T_RU_SAE), T_rz_sae(EvexModifierRounding::T_RZ_SAE)
 		, T_z()
 #ifdef XBYAK64
 		, rax(Operand::RAX), rcx(Operand::RCX), rdx(Operand::RDX), rbx(Operand::RBX), rsp(Operand::RSP), rbp(Operand::RBP), rsi(Operand::RSI), rdi(Operand::RDI), r8(Operand::R8), r9(Operand::R9), r10(Operand::R10), r11(Operand::R11), r12(Operand::R12), r13(Operand::R13), r14(Operand::R14), r15(Operand::R15)
@@ -2329,6 +2799,7 @@ public:
 		, zmm8(8), zmm9(9), zmm10(10), zmm11(11), zmm12(12), zmm13(13), zmm14(14), zmm15(15)
 		, zmm16(16), zmm17(17), zmm18(18), zmm19(19), zmm20(20), zmm21(21), zmm22(22), zmm23(23)
 		, zmm24(24), zmm25(25), zmm26(26), zmm27(27), zmm28(28), zmm29(29), zmm30(30), zmm31(31)
+		, tmm0(0), tmm1(1), tmm2(2), tmm3(3), tmm4(4), tmm5(5), tmm6(6), tmm7(7)
 		// for my convenience
 		, xm8(xmm8), xm9(xmm9), xm10(xmm10), xm11(xmm11), xm12(xmm12), xm13(xmm13), xm14(xmm14), xm15(xmm15)
 		, xm16(xmm16), xm17(xmm17), xm18(xmm18), xm19(xmm19), xm20(xmm20), xm21(xmm21), xm22(xmm22), xm23(xmm23)
@@ -2344,11 +2815,14 @@ public:
 #ifndef XBYAK_DISABLE_SEGMENT
 		, es(Segment::es), cs(Segment::cs), ss(Segment::ss), ds(Segment::ds), fs(Segment::fs), gs(Segment::gs)
 #endif
+		, isDefaultJmpNEAR_(false)
+		, defaultEncoding_(EvexEncoding)
 	{
 		labelMgr_.set(this);
 	}
 	void reset()
 	{
+		ClearError();
 		resetSize();
 		labelMgr_.reset();
 		labelMgr_.set(this);
@@ -2358,11 +2832,16 @@ public:
 		MUST call ready() to complete generating code if you use AutoGrow mode.
 		It is not necessary for the other mode if hasUndefinedLabel() is true.
 	*/
-	void ready()
+	void ready(ProtectMode mode = PROTECT_RWE)
 	{
-		if (hasUndefinedLabel()) throw Error(ERR_LABEL_IS_NOT_FOUND);
-		if (isAutoGrow()) calcJmpAddress();
+		if (hasUndefinedLabel()) XBYAK_THROW(ERR_LABEL_IS_NOT_FOUND)
+		if (isAutoGrow()) {
+			calcJmpAddress();
+			if (useProtect()) setProtectMode(mode);
+		}
 	}
+	// set read/exec
+	void readyRE() { return ready(PROTECT_RE); }
 #ifdef XBYAK_TEST
 	void dump(bool doClear = true)
 	{
@@ -2371,49 +2850,110 @@ public:
 	}
 #endif
 
+#ifdef XBYAK_UNDEF_JNL
+	#undef jnl
+#endif
+
+	// set default encoding to select Vex or Evex
+	void setDefaultEncoding(PreferredEncoding encoding) { defaultEncoding_ = encoding; }
+
+	/*
+		use single byte nop if useMultiByteNop = false
+	*/
+	void nop(size_t size = 1, bool useMultiByteNop = true)
+	{
+		if (!useMultiByteNop) {
+			for (size_t i = 0; i < size; i++) {
+				db(0x90);
+			}
+			return;
+		}
+		/*
+			Intel Architectures Software Developer's Manual Volume 2
+			recommended multi-byte sequence of NOP instruction
+			AMD and Intel seem to agree on the same sequences for up to 9 bytes:
+			https://support.amd.com/TechDocs/55723_SOG_Fam_17h_Processors_3.00.pdf
+		*/
+		static const uint8_t nopTbl[9][9] = {
+			{0x90},
+			{0x66, 0x90},
+			{0x0F, 0x1F, 0x00},
+			{0x0F, 0x1F, 0x40, 0x00},
+			{0x0F, 0x1F, 0x44, 0x00, 0x00},
+			{0x66, 0x0F, 0x1F, 0x44, 0x00, 0x00},
+			{0x0F, 0x1F, 0x80, 0x00, 0x00, 0x00, 0x00},
+			{0x0F, 0x1F, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00},
+			{0x66, 0x0F, 0x1F, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00},
+		};
+		const size_t n = sizeof(nopTbl) / sizeof(nopTbl[0]);
+		while (size > 0) {
+			size_t len = (std::min)(n, size);
+			const uint8_t *seq = nopTbl[len - 1];
+			db(seq, len);
+			size -= len;
+		}
+	}
+
 #ifndef XBYAK_DONT_READ_LIST
 #include "xbyak_mnemonic.h"
-	void align(int x = 16)
+	/*
+		use single byte nop if useMultiByteNop = false
+	*/
+	void align(size_t x = 16, bool useMultiByteNop = true)
 	{
 		if (x == 1) return;
-		if (x < 1 || (x & (x - 1))) throw Error(ERR_BAD_ALIGN);
-		if (isAutoGrow() && x > (int)inner::ALIGN_PAGE_SIZE) fprintf(stderr, "warning:autoGrow mode does not support %d align\n", x);
-		while (size_t(getCurr()) % x) {
-			nop();
+		if (x < 1 || (x & (x - 1))) XBYAK_THROW(ERR_BAD_ALIGN)
+		if (isAutoGrow()) XBYAK_THROW(ERR_BAD_ALIGN)
+		size_t remain = size_t(getCurr()) % x;
+		if (remain) {
+			nop(x - remain, useMultiByteNop);
 		}
 	}
 #endif
 };
 
+template <>
+inline void CodeGenerator::mov(const NativeReg& reg, const char *label) // can't use std::string
+{
+	assert(label);
+	mov_imm(reg, dummyAddr);
+	putL(label);
+}
+
 namespace util {
-static const Mmx mm0(0), mm1(1), mm2(2), mm3(3), mm4(4), mm5(5), mm6(6), mm7(7);
-static const Xmm xmm0(0), xmm1(1), xmm2(2), xmm3(3), xmm4(4), xmm5(5), xmm6(6), xmm7(7);
-static const Ymm ymm0(0), ymm1(1), ymm2(2), ymm3(3), ymm4(4), ymm5(5), ymm6(6), ymm7(7);
-static const Zmm zmm0(0), zmm1(1), zmm2(2), zmm3(3), zmm4(4), zmm5(5), zmm6(6), zmm7(7);
-static const Reg32 eax(Operand::EAX), ecx(Operand::ECX), edx(Operand::EDX), ebx(Operand::EBX), esp(Operand::ESP), ebp(Operand::EBP), esi(Operand::ESI), edi(Operand::EDI);
-static const Reg16 ax(Operand::AX), cx(Operand::CX), dx(Operand::DX), bx(Operand::BX), sp(Operand::SP), bp(Operand::BP), si(Operand::SI), di(Operand::DI);
-static const Reg8 al(Operand::AL), cl(Operand::CL), dl(Operand::DL), bl(Operand::BL), ah(Operand::AH), ch(Operand::CH), dh(Operand::DH), bh(Operand::BH);
-static const AddressFrame ptr(0), byte(8), word(16), dword(32), qword(64);
-static const Fpu st0(0), st1(1), st2(2), st3(3), st4(4), st5(5), st6(6), st7(7);
-static const Opmask k0(0), k1(1), k2(2), k3(3), k4(4), k5(5), k6(6), k7(7);
+static const XBYAK_CONSTEXPR Mmx mm0(0), mm1(1), mm2(2), mm3(3), mm4(4), mm5(5), mm6(6), mm7(7);
+static const XBYAK_CONSTEXPR Xmm xmm0(0), xmm1(1), xmm2(2), xmm3(3), xmm4(4), xmm5(5), xmm6(6), xmm7(7);
+static const XBYAK_CONSTEXPR Ymm ymm0(0), ymm1(1), ymm2(2), ymm3(3), ymm4(4), ymm5(5), ymm6(6), ymm7(7);
+static const XBYAK_CONSTEXPR Zmm zmm0(0), zmm1(1), zmm2(2), zmm3(3), zmm4(4), zmm5(5), zmm6(6), zmm7(7);
+static const XBYAK_CONSTEXPR Reg32 eax(Operand::EAX), ecx(Operand::ECX), edx(Operand::EDX), ebx(Operand::EBX), esp(Operand::ESP), ebp(Operand::EBP), esi(Operand::ESI), edi(Operand::EDI);
+static const XBYAK_CONSTEXPR Reg16 ax(Operand::AX), cx(Operand::CX), dx(Operand::DX), bx(Operand::BX), sp(Operand::SP), bp(Operand::BP), si(Operand::SI), di(Operand::DI);
+static const XBYAK_CONSTEXPR Reg8 al(Operand::AL), cl(Operand::CL), dl(Operand::DL), bl(Operand::BL), ah(Operand::AH), ch(Operand::CH), dh(Operand::DH), bh(Operand::BH);
+static const XBYAK_CONSTEXPR AddressFrame ptr(0), byte(8), word(16), dword(32), qword(64), xword(128), yword(256), zword(512);
+static const XBYAK_CONSTEXPR AddressFrame ptr_b(0, true), xword_b(128, true), yword_b(256, true), zword_b(512, true);
+static const XBYAK_CONSTEXPR Fpu st0(0), st1(1), st2(2), st3(3), st4(4), st5(5), st6(6), st7(7);
+static const XBYAK_CONSTEXPR Opmask k0(0), k1(1), k2(2), k3(3), k4(4), k5(5), k6(6), k7(7);
+static const XBYAK_CONSTEXPR BoundsReg bnd0(0), bnd1(1), bnd2(2), bnd3(3);
+static const XBYAK_CONSTEXPR EvexModifierRounding T_sae(EvexModifierRounding::T_SAE), T_rn_sae(EvexModifierRounding::T_RN_SAE), T_rd_sae(EvexModifierRounding::T_RD_SAE), T_ru_sae(EvexModifierRounding::T_RU_SAE), T_rz_sae(EvexModifierRounding::T_RZ_SAE);
+static const XBYAK_CONSTEXPR EvexModifierZero T_z;
 #ifdef XBYAK64
-static const Reg64 rax(Operand::RAX), rcx(Operand::RCX), rdx(Operand::RDX), rbx(Operand::RBX), rsp(Operand::RSP), rbp(Operand::RBP), rsi(Operand::RSI), rdi(Operand::RDI), r8(Operand::R8), r9(Operand::R9), r10(Operand::R10), r11(Operand::R11), r12(Operand::R12), r13(Operand::R13), r14(Operand::R14), r15(Operand::R15);
-static const Reg32 r8d(8), r9d(9), r10d(10), r11d(11), r12d(12), r13d(13), r14d(14), r15d(15);
-static const Reg16 r8w(8), r9w(9), r10w(10), r11w(11), r12w(12), r13w(13), r14w(14), r15w(15);
-static const Reg8 r8b(8), r9b(9), r10b(10), r11b(11), r12b(12), r13b(13), r14b(14), r15b(15), spl(Operand::SPL, true), bpl(Operand::BPL, true), sil(Operand::SIL, true), dil(Operand::DIL, true);
-static const Xmm xmm8(8), xmm9(9), xmm10(10), xmm11(11), xmm12(12), xmm13(13), xmm14(14), xmm15(15);
-static const Xmm xmm16(16), xmm17(17), xmm18(18), xmm19(19), xmm20(20), xmm21(21), xmm22(22), xmm23(23);
-static const Xmm xmm24(24), xmm25(25), xmm26(26), xmm27(27), xmm28(28), xmm29(29), xmm30(30), xmm31(31);
-static const Ymm ymm8(8), ymm9(9), ymm10(10), ymm11(11), ymm12(12), ymm13(13), ymm14(14), ymm15(15);
-static const Ymm ymm16(16), ymm17(17), ymm18(18), ymm19(19), ymm20(20), ymm21(21), ymm22(22), ymm23(23);
-static const Ymm ymm24(24), ymm25(25), ymm26(26), ymm27(27), ymm28(28), ymm29(29), ymm30(30), ymm31(31);
-static const Zmm zmm8(8), zmm9(9), zmm10(10), zmm11(11), zmm12(12), zmm13(13), zmm14(14), zmm15(15);
-static const Zmm zmm16(16), zmm17(17), zmm18(18), zmm19(19), zmm20(20), zmm21(21), zmm22(22), zmm23(23);
-static const Zmm zmm24(24), zmm25(25), zmm26(26), zmm27(27), zmm28(28), zmm29(29), zmm30(30), zmm31(31);
-static const RegRip rip;
+static const XBYAK_CONSTEXPR Reg64 rax(Operand::RAX), rcx(Operand::RCX), rdx(Operand::RDX), rbx(Operand::RBX), rsp(Operand::RSP), rbp(Operand::RBP), rsi(Operand::RSI), rdi(Operand::RDI), r8(Operand::R8), r9(Operand::R9), r10(Operand::R10), r11(Operand::R11), r12(Operand::R12), r13(Operand::R13), r14(Operand::R14), r15(Operand::R15);
+static const XBYAK_CONSTEXPR Reg32 r8d(8), r9d(9), r10d(10), r11d(11), r12d(12), r13d(13), r14d(14), r15d(15);
+static const XBYAK_CONSTEXPR Reg16 r8w(8), r9w(9), r10w(10), r11w(11), r12w(12), r13w(13), r14w(14), r15w(15);
+static const XBYAK_CONSTEXPR Reg8 r8b(8), r9b(9), r10b(10), r11b(11), r12b(12), r13b(13), r14b(14), r15b(15), spl(Operand::SPL, true), bpl(Operand::BPL, true), sil(Operand::SIL, true), dil(Operand::DIL, true);
+static const XBYAK_CONSTEXPR Xmm xmm8(8), xmm9(9), xmm10(10), xmm11(11), xmm12(12), xmm13(13), xmm14(14), xmm15(15);
+static const XBYAK_CONSTEXPR Xmm xmm16(16), xmm17(17), xmm18(18), xmm19(19), xmm20(20), xmm21(21), xmm22(22), xmm23(23);
+static const XBYAK_CONSTEXPR Xmm xmm24(24), xmm25(25), xmm26(26), xmm27(27), xmm28(28), xmm29(29), xmm30(30), xmm31(31);
+static const XBYAK_CONSTEXPR Ymm ymm8(8), ymm9(9), ymm10(10), ymm11(11), ymm12(12), ymm13(13), ymm14(14), ymm15(15);
+static const XBYAK_CONSTEXPR Ymm ymm16(16), ymm17(17), ymm18(18), ymm19(19), ymm20(20), ymm21(21), ymm22(22), ymm23(23);
+static const XBYAK_CONSTEXPR Ymm ymm24(24), ymm25(25), ymm26(26), ymm27(27), ymm28(28), ymm29(29), ymm30(30), ymm31(31);
+static const XBYAK_CONSTEXPR Zmm zmm8(8), zmm9(9), zmm10(10), zmm11(11), zmm12(12), zmm13(13), zmm14(14), zmm15(15);
+static const XBYAK_CONSTEXPR Zmm zmm16(16), zmm17(17), zmm18(18), zmm19(19), zmm20(20), zmm21(21), zmm22(22), zmm23(23);
+static const XBYAK_CONSTEXPR Zmm zmm24(24), zmm25(25), zmm26(26), zmm27(27), zmm28(28), zmm29(29), zmm30(30), zmm31(31);
+static const XBYAK_CONSTEXPR Zmm tmm0(0), tmm1(1), tmm2(2), tmm3(3), tmm4(4), tmm5(5), tmm6(6), tmm7(7);
+static const XBYAK_CONSTEXPR RegRip rip;
 #endif
 #ifndef XBYAK_DISABLE_SEGMENT
-static const Segment es(Segment::es), cs(Segment::cs), ss(Segment::ss), ds(Segment::ds), fs(Segment::fs), gs(Segment::gs);
+static const XBYAK_CONSTEXPR Segment es(Segment::es), cs(Segment::cs), ss(Segment::ss), ds(Segment::ds), fs(Segment::fs), gs(Segment::gs);
 #endif
 } // util
 
@@ -2421,6 +2961,10 @@ static const Segment es(Segment::es), cs(Segment::cs), ss(Segment::ss), ds(Segme
 	#pragma warning(pop)
 #endif
 
+#if defined(__GNUC__) && !defined(__clang__)
+	#pragma GCC diagnostic pop
+#endif
+
 } // end of namespace
 
 #endif // XBYAK_XBYAK_H_
diff --git a/dep/xbyak/xbyak/xbyak_mnemonic.h b/dep/xbyak/xbyak/xbyak_mnemonic.h
index 175932fec..edc76c5bb 100644
--- a/dep/xbyak/xbyak/xbyak_mnemonic.h
+++ b/dep/xbyak/xbyak/xbyak_mnemonic.h
@@ -1,8 +1,10 @@
-const char *getVersionString() const { return "5.41"; }
-void adc(const Operand& op, uint32 imm) { opRM_I(op, imm, 0x10, 2); }
+const char *getVersionString() const { return "6.73"; }
+void aadd(const Address& addr, const Reg32e &reg) { opModM(addr, reg, 0x0F, 0x38, 0x0FC); }
+void aand(const Address& addr, const Reg32e &reg) { db(0x66); opModM(addr, reg, 0x0F, 0x38, 0x0FC); }
+void adc(const Operand& op, uint32_t imm) { opRM_I(op, imm, 0x10, 2); }
 void adc(const Operand& op1, const Operand& op2) { opRM_RM(op1, op2, 0x10); }
 void adcx(const Reg32e& reg, const Operand& op) { opGen(reg, op, 0xF6, 0x66, isREG32_REG32orMEM, NONE, 0x38); }
-void add(const Operand& op, uint32 imm) { opRM_I(op, imm, 0x00, 0); }
+void add(const Operand& op, uint32_t imm) { opRM_I(op, imm, 0x00, 0); }
 void add(const Operand& op1, const Operand& op2) { opRM_RM(op1, op2, 0x00); }
 void addpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x58, 0x66, isXMM_XMMorMEM); }
 void addps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x58, 0x100, isXMM_XMMorMEM); }
@@ -16,72 +18,87 @@ void aesdeclast(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0xDF, 0x66,
 void aesenc(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0xDC, 0x66, isXMM_XMMorMEM, NONE, 0x38); }
 void aesenclast(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0xDD, 0x66, isXMM_XMMorMEM, NONE, 0x38); }
 void aesimc(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0xDB, 0x66, isXMM_XMMorMEM, NONE, 0x38); }
-void aeskeygenassist(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0xDF, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
-void and_(const Operand& op, uint32 imm) { opRM_I(op, imm, 0x20, 4); }
+void aeskeygenassist(const Xmm& xmm, const Operand& op, uint8_t imm) { opGen(xmm, op, 0xDF, 0x66, isXMM_XMMorMEM, imm, 0x3A); }
+void and_(const Operand& op, uint32_t imm) { opRM_I(op, imm, 0x20, 4); }
 void and_(const Operand& op1, const Operand& op2) { opRM_RM(op1, op2, 0x20); }
 void andn(const Reg32e& r1, const Reg32e& r2, const Operand& op) { opGpr(r1, r2, op, T_0F38, 0xf2, true); }
 void andnpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x55, 0x66, isXMM_XMMorMEM); }
 void andnps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x55, 0x100, isXMM_XMMorMEM); }
 void andpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x54, 0x66, isXMM_XMMorMEM); }
 void andps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x54, 0x100, isXMM_XMMorMEM); }
+void aor(const Address& addr, const Reg32e &reg) { db(0xF2); opModM(addr, reg, 0x0F, 0x38, 0x0FC); }
+void axor(const Address& addr, const Reg32e &reg) { db(0xF3); opModM(addr, reg, 0x0F, 0x38, 0x0FC); }
 void bextr(const Reg32e& r1, const Operand& op, const Reg32e& r2) { opGpr(r1, op, r2, T_0F38, 0xf7, false); }
-void blendpd(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x0D, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
-void blendps(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x0C, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
+void blendpd(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x0D, 0x66, isXMM_XMMorMEM, static_cast<uint8_t>(imm), 0x3A); }
+void blendps(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x0C, 0x66, isXMM_XMMorMEM, static_cast<uint8_t>(imm), 0x3A); }
 void blendvpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x15, 0x66, isXMM_XMMorMEM, NONE, 0x38); }
 void blendvps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x14, 0x66, isXMM_XMMorMEM, NONE, 0x38); }
 void blsi(const Reg32e& r, const Operand& op) { opGpr(Reg32e(3, r.getBit()), op, r, T_0F38, 0xf3, false); }
 void blsmsk(const Reg32e& r, const Operand& op) { opGpr(Reg32e(2, r.getBit()), op, r, T_0F38, 0xf3, false); }
 void blsr(const Reg32e& r, const Operand& op) { opGpr(Reg32e(1, r.getBit()), op, r, T_0F38, 0xf3, false); }
+void bnd() { db(0xF2); }
+void bndcl(const BoundsReg& bnd, const Operand& op) { db(0xF3); opR_ModM(op, i32e, bnd.getIdx(), 0x0F, 0x1A, NONE, !op.isMEM()); }
+void bndcn(const BoundsReg& bnd, const Operand& op) { db(0xF2); opR_ModM(op, i32e, bnd.getIdx(), 0x0F, 0x1B, NONE, !op.isMEM()); }
+void bndcu(const BoundsReg& bnd, const Operand& op) { db(0xF2); opR_ModM(op, i32e, bnd.getIdx(), 0x0F, 0x1A, NONE, !op.isMEM()); }
+void bndldx(const BoundsReg& bnd, const Address& addr) { opMIB(addr, bnd, 0x0F, 0x1A); }
+void bndmk(const BoundsReg& bnd, const Address& addr) { db(0xF3); opModM(addr, bnd, 0x0F, 0x1B); }
+void bndmov(const Address& addr, const BoundsReg& bnd) { db(0x66); opModM(addr, bnd, 0x0F, 0x1B); }
+void bndmov(const BoundsReg& bnd, const Operand& op) { db(0x66); opModRM(bnd, op, op.isBNDREG(), op.isMEM(), 0x0F, 0x1A); }
+void bndstx(const Address& addr, const BoundsReg& bnd) { opMIB(addr, bnd, 0x0F, 0x1B); }
 void bsf(const Reg&reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0xBC); }
 void bsr(const Reg&reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0xBD); }
 void bswap(const Reg32e& reg) { opModR(Reg32(1), reg, 0x0F); }
 void bt(const Operand& op, const Reg& reg) { opModRM(reg, op, op.isREG(16|32|64) && op.getBit() == reg.getBit(), op.isMEM(), 0x0f, 0xA3); }
-void bt(const Operand& op, uint8 imm) { opR_ModM(op, 16|32|64, 4, 0x0f, 0xba, NONE, false, 1); db(imm); }
+void bt(const Operand& op, uint8_t imm) { opR_ModM(op, 16|32|64, 4, 0x0f, 0xba, NONE, false, 1); db(imm); }
 void btc(const Operand& op, const Reg& reg) { opModRM(reg, op, op.isREG(16|32|64) && op.getBit() == reg.getBit(), op.isMEM(), 0x0f, 0xBB); }
-void btc(const Operand& op, uint8 imm) { opR_ModM(op, 16|32|64, 7, 0x0f, 0xba, NONE, false, 1); db(imm); }
+void btc(const Operand& op, uint8_t imm) { opR_ModM(op, 16|32|64, 7, 0x0f, 0xba, NONE, false, 1); db(imm); }
 void btr(const Operand& op, const Reg& reg) { opModRM(reg, op, op.isREG(16|32|64) && op.getBit() == reg.getBit(), op.isMEM(), 0x0f, 0xB3); }
-void btr(const Operand& op, uint8 imm) { opR_ModM(op, 16|32|64, 6, 0x0f, 0xba, NONE, false, 1); db(imm); }
+void btr(const Operand& op, uint8_t imm) { opR_ModM(op, 16|32|64, 6, 0x0f, 0xba, NONE, false, 1); db(imm); }
 void bts(const Operand& op, const Reg& reg) { opModRM(reg, op, op.isREG(16|32|64) && op.getBit() == reg.getBit(), op.isMEM(), 0x0f, 0xAB); }
-void bts(const Operand& op, uint8 imm) { opR_ModM(op, 16|32|64, 5, 0x0f, 0xba, NONE, false, 1); db(imm); }
+void bts(const Operand& op, uint8_t imm) { opR_ModM(op, 16|32|64, 5, 0x0f, 0xba, NONE, false, 1); db(imm); }
 void bzhi(const Reg32e& r1, const Operand& op, const Reg32e& r2) { opGpr(r1, op, r2, T_0F38, 0xf5, false); }
 void cbw() { db(0x66); db(0x98); }
 void cdq() { db(0x99); }
 void clc() { db(0xF8); }
 void cld() { db(0xFC); }
+void cldemote(const Address& addr) { opMIB(addr, eax, 0x0F, 0x1C); }
 void clflush(const Address& addr) { opModM(addr, Reg32(7), 0x0F, 0xAE); }
+void clflushopt(const Address& addr) { db(0x66); opModM(addr, Reg32(7), 0x0F, 0xAE); }
 void cli() { db(0xFA); }
+void clwb(const Address& addr) { db(0x66); opMIB(addr, esi, 0x0F, 0xAE); }
+void clzero() { db(0x0F); db(0x01); db(0xFC); }
 void cmc() { db(0xF5); }
-void cmova(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 7); }
-void cmovae(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 3); }
-void cmovb(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 2); }
-void cmovbe(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 6); }
-void cmovc(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 2); }
-void cmove(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 4); }
-void cmovg(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 15); }
-void cmovge(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 13); }
-void cmovl(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 12); }
-void cmovle(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 14); }
-void cmovna(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 6); }
-void cmovnae(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 2); }
-void cmovnb(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 3); }
-void cmovnbe(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 7); }
-void cmovnc(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 3); }
-void cmovne(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 5); }
-void cmovng(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 14); }
-void cmovnge(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 12); }
-void cmovnl(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 13); }
-void cmovnle(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 15); }
-void cmovno(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 1); }
-void cmovnp(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 11); }
-void cmovns(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 9); }
-void cmovnz(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 5); }
-void cmovo(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 0); }
-void cmovp(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 10); }
-void cmovpe(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 10); }
-void cmovpo(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 11); }
-void cmovs(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 8); }
-void cmovz(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 4); }
-void cmp(const Operand& op, uint32 imm) { opRM_I(op, imm, 0x38, 7); }
+void cmova(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 7); }//-V524
+void cmovae(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 3); }//-V524
+void cmovb(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 2); }//-V524
+void cmovbe(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 6); }//-V524
+void cmovc(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 2); }//-V524
+void cmove(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 4); }//-V524
+void cmovg(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 15); }//-V524
+void cmovge(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 13); }//-V524
+void cmovl(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 12); }//-V524
+void cmovle(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 14); }//-V524
+void cmovna(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 6); }//-V524
+void cmovnae(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 2); }//-V524
+void cmovnb(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 3); }//-V524
+void cmovnbe(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 7); }//-V524
+void cmovnc(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 3); }//-V524
+void cmovne(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 5); }//-V524
+void cmovng(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 14); }//-V524
+void cmovnge(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 12); }//-V524
+void cmovnl(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 13); }//-V524
+void cmovnle(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 15); }//-V524
+void cmovno(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 1); }//-V524
+void cmovnp(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 11); }//-V524
+void cmovns(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 9); }//-V524
+void cmovnz(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 5); }//-V524
+void cmovo(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 0); }//-V524
+void cmovp(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 10); }//-V524
+void cmovpe(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 10); }//-V524
+void cmovpo(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 11); }//-V524
+void cmovs(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 8); }//-V524
+void cmovz(const Reg& reg, const Operand& op) { opModRM(reg, op, op.isREG(16 | i32e), op.isMEM(), 0x0F, 0x40 | 4); }//-V524
+void cmp(const Operand& op, uint32_t imm) { opRM_I(op, imm, 0x38, 7); }
 void cmp(const Operand& op1, const Operand& op2) { opRM_RM(op1, op2, 0x38); }
 void cmpeqpd(const Xmm& x, const Operand& op) { cmppd(x, op, 0); }
 void cmpeqps(const Xmm& x, const Operand& op) { cmpps(x, op, 0); }
@@ -111,10 +128,13 @@ void cmpordpd(const Xmm& x, const Operand& op) { cmppd(x, op, 7); }
 void cmpordps(const Xmm& x, const Operand& op) { cmpps(x, op, 7); }
 void cmpordsd(const Xmm& x, const Operand& op) { cmpsd(x, op, 7); }
 void cmpordss(const Xmm& x, const Operand& op) { cmpss(x, op, 7); }
-void cmppd(const Xmm& xmm, const Operand& op, uint8 imm8) { opGen(xmm, op, 0xC2, 0x66, isXMM_XMMorMEM, imm8); }
-void cmpps(const Xmm& xmm, const Operand& op, uint8 imm8) { opGen(xmm, op, 0xC2, 0x100, isXMM_XMMorMEM, imm8); }
-void cmpsd(const Xmm& xmm, const Operand& op, uint8 imm8) { opGen(xmm, op, 0xC2, 0xF2, isXMM_XMMorMEM, imm8); }
-void cmpss(const Xmm& xmm, const Operand& op, uint8 imm8) { opGen(xmm, op, 0xC2, 0xF3, isXMM_XMMorMEM, imm8); }
+void cmppd(const Xmm& xmm, const Operand& op, uint8_t imm8) { opGen(xmm, op, 0xC2, 0x66, isXMM_XMMorMEM, imm8); }
+void cmpps(const Xmm& xmm, const Operand& op, uint8_t imm8) { opGen(xmm, op, 0xC2, 0x100, isXMM_XMMorMEM, imm8); }
+void cmpsb() { db(0xA6); }
+void cmpsd() { db(0xA7); }
+void cmpsd(const Xmm& xmm, const Operand& op, uint8_t imm8) { opGen(xmm, op, 0xC2, 0xF2, isXMM_XMMorMEM, imm8); }
+void cmpss(const Xmm& xmm, const Operand& op, uint8_t imm8) { opGen(xmm, op, 0xC2, 0xF3, isXMM_XMMorMEM, imm8); }
+void cmpsw() { db(0x66); db(0xA7); }
 void cmpunordpd(const Xmm& x, const Operand& op) { cmppd(x, op, 3); }
 void cmpunordps(const Xmm& x, const Operand& op) { cmpps(x, op, 3); }
 void cmpunordsd(const Xmm& x, const Operand& op) { cmpsd(x, op, 3); }
@@ -155,10 +175,13 @@ void divpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x5E, 0x66, isXMM
 void divps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x5E, 0x100, isXMM_XMMorMEM); }
 void divsd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x5E, 0xF2, isXMM_XMMorMEM); }
 void divss(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x5E, 0xF3, isXMM_XMMorMEM); }
-void dppd(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x41, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
-void dpps(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x40, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
+void dppd(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x41, 0x66, isXMM_XMMorMEM, static_cast<uint8_t>(imm), 0x3A); }
+void dpps(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x40, 0x66, isXMM_XMMorMEM, static_cast<uint8_t>(imm), 0x3A); }
 void emms() { db(0x0F); db(0x77); }
-void extractps(const Operand& op, const Xmm& xmm, uint8 imm) { opExt(op, xmm, 0x17, imm); }
+void endbr32() { db(0xF3); db(0x0F); db(0x1E); db(0xFB); }
+void endbr64() { db(0xF3); db(0x0F); db(0x1E); db(0xFA); }
+void enter(uint16_t x, uint8_t y) { db(0xC8); dw(x); db(y); }
+void extractps(const Operand& op, const Xmm& xmm, uint8_t imm) { opExt(op, xmm, 0x17, imm); }
 void f2xm1() { db(0xD9); db(0xF0); }
 void fabs() { db(0xD9); db(0xE1); }
 void fadd(const Address& addr) { opFpuMem(addr, 0x00, 0xD8, 0xDC, 0, 0); }
@@ -167,7 +190,10 @@ void fadd(const Fpu& reg1, const Fpu& reg2) { opFpuFpu(reg1, reg2, 0xD8C0, 0xDCC
 void faddp() { db(0xDE); db(0xC1); }
 void faddp(const Fpu& reg1) { opFpuFpu(reg1, st0, 0x0000, 0xDEC0); }
 void faddp(const Fpu& reg1, const Fpu& reg2) { opFpuFpu(reg1, reg2, 0x0000, 0xDEC0); }
+void fbld(const Address& addr) { opModM(addr, Reg32(4), 0xDF, 0x100); }
+void fbstp(const Address& addr) { opModM(addr, Reg32(6), 0xDF, 0x100); }
 void fchs() { db(0xD9); db(0xE0); }
+void fclex() { db(0x9B); db(0xDB); db(0xE2); }
 void fcmovb(const Fpu& reg1) { opFpuFpu(st0, reg1, 0xDAC0, 0x00C0); }
 void fcmovb(const Fpu& reg1, const Fpu& reg2) { opFpuFpu(reg1, reg2, 0xDAC0, 0x00C0); }
 void fcmovbe(const Fpu& reg1) { opFpuFpu(st0, reg1, 0xDAD0, 0x00D0); }
@@ -228,6 +254,7 @@ void fld(const Address& addr) { opFpuMem(addr, 0x00, 0xD9, 0xDD, 0, 0); }
 void fld(const Fpu& reg) { opFpu(reg, 0xD9, 0xC0); }
 void fld1() { db(0xD9); db(0xE8); }
 void fldcw(const Address& addr) { opModM(addr, Reg32(5), 0xD9, 0x100); }
+void fldenv(const Address& addr) { opModM(addr, Reg32(4), 0xD9, 0x100); }
 void fldl2e() { db(0xD9); db(0xEA); }
 void fldl2t() { db(0xD9); db(0xE9); }
 void fldlg2() { db(0xD9); db(0xEC); }
@@ -240,22 +267,33 @@ void fmul(const Fpu& reg1, const Fpu& reg2) { opFpuFpu(reg1, reg2, 0xD8C8, 0xDCC
 void fmulp() { db(0xDE); db(0xC9); }
 void fmulp(const Fpu& reg1) { opFpuFpu(reg1, st0, 0x0000, 0xDEC8); }
 void fmulp(const Fpu& reg1, const Fpu& reg2) { opFpuFpu(reg1, reg2, 0x0000, 0xDEC8); }
+void fnclex() { db(0xDB); db(0xE2); }
 void fninit() { db(0xDB); db(0xE3); }
 void fnop() { db(0xD9); db(0xD0); }
+void fnsave(const Address& addr) { opModM(addr, Reg32(6), 0xDD, 0x100); }
+void fnstcw(const Address& addr) { opModM(addr, Reg32(7), 0xD9, 0x100); }
+void fnstenv(const Address& addr) { opModM(addr, Reg32(6), 0xD9, 0x100); }
+void fnstsw(const Address& addr) { opModM(addr, Reg32(7), 0xDD, 0x100); }
+void fnstsw(const Reg16& r) { if (r.getIdx() != Operand::AX) XBYAK_THROW(ERR_BAD_PARAMETER) db(0xDF); db(0xE0); }
 void fpatan() { db(0xD9); db(0xF3); }
 void fprem() { db(0xD9); db(0xF8); }
 void fprem1() { db(0xD9); db(0xF5); }
 void fptan() { db(0xD9); db(0xF2); }
 void frndint() { db(0xD9); db(0xFC); }
+void frstor(const Address& addr) { opModM(addr, Reg32(4), 0xDD, 0x100); }
+void fsave(const Address& addr) { db(0x9B); opModM(addr, Reg32(6), 0xDD, 0x100); }
 void fscale() { db(0xD9); db(0xFD); }
 void fsin() { db(0xD9); db(0xFE); }
 void fsincos() { db(0xD9); db(0xFB); }
 void fsqrt() { db(0xD9); db(0xFA); }
 void fst(const Address& addr) { opFpuMem(addr, 0x00, 0xD9, 0xDD, 2, 0); }
 void fst(const Fpu& reg) { opFpu(reg, 0xDD, 0xD0); }
-void fstcw(const Address& addr) { db(0x9B); opModM(addr, Reg32(7), 0xD9, NONE); }
+void fstcw(const Address& addr) { db(0x9B); opModM(addr, Reg32(7), 0xD9, 0x100); }
+void fstenv(const Address& addr) { db(0x9B); opModM(addr, Reg32(6), 0xD9, 0x100); }
 void fstp(const Address& addr) { opFpuMem(addr, 0x00, 0xD9, 0xDD, 3, 0); }
 void fstp(const Fpu& reg) { opFpu(reg, 0xDD, 0xD8); }
+void fstsw(const Address& addr) { db(0x9B); opModM(addr, Reg32(7), 0xDD, 0x100); }
+void fstsw(const Reg16& r) { if (r.getIdx() != Operand::AX) XBYAK_THROW(ERR_BAD_PARAMETER) db(0x9B); db(0xDF); db(0xE0); }
 void fsub(const Address& addr) { opFpuMem(addr, 0x00, 0xD8, 0xDC, 4, 0); }
 void fsub(const Fpu& reg1) { opFpuFpu(st0, reg1, 0xD8E0, 0xDCE8); }
 void fsub(const Fpu& reg1, const Fpu& reg2) { opFpuFpu(reg1, reg2, 0xD8E0, 0xDCE8); }
@@ -282,146 +320,171 @@ void fwait() { db(0x9B); }
 void fxam() { db(0xD9); db(0xE5); }
 void fxch() { db(0xD9); db(0xC9); }
 void fxch(const Fpu& reg) { opFpu(reg, 0xD9, 0xC8); }
+void fxrstor(const Address& addr) { opModM(addr, Reg32(1), 0x0F, 0xAE); }
 void fxtract() { db(0xD9); db(0xF4); }
 void fyl2x() { db(0xD9); db(0xF1); }
 void fyl2xp1() { db(0xD9); db(0xF9); }
+void gf2p8affineinvqb(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0xCF, 0x66, isXMM_XMMorMEM, static_cast<uint8_t>(imm), 0x3A); }
+void gf2p8affineqb(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0xCE, 0x66, isXMM_XMMorMEM, static_cast<uint8_t>(imm), 0x3A); }
+void gf2p8mulb(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0xCF, 0x66, isXMM_XMMorMEM, NONE, 0x38); }
 void haddpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x7C, 0x66, isXMM_XMMorMEM); }
 void haddps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x7C, 0xF2, isXMM_XMMorMEM); }
+void hlt() { db(0xF4); }
 void hsubpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x7D, 0x66, isXMM_XMMorMEM); }
 void hsubps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x7D, 0xF2, isXMM_XMMorMEM); }
 void idiv(const Operand& op) { opR_ModM(op, 0, 7, 0xF6); }
 void imul(const Operand& op) { opR_ModM(op, 0, 5, 0xF6); }
+void in_(const Reg& a, const Reg& d) { opInOut(a, d, 0xEC); }
+void in_(const Reg& a, uint8_t v) { opInOut(a, 0xE4, v); }
 void inc(const Operand& op) { opIncDec(op, 0x40, 0); }
-void insertps(const Xmm& xmm, const Operand& op, uint8 imm) { opGen(xmm, op, 0x21, 0x66, isXMM_XMMorMEM, imm, 0x3A); }
-void ja(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x77, 0x87, 0x0F); }
-void ja(const char *label, LabelType type = T_AUTO) { ja(std::string(label), type); }
-void ja(const void *addr) { opJmpAbs(addr, T_NEAR, 0x77, 0x87, 0x0F); }
-void ja(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x77, 0x87, 0x0F); }
-void jae(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x73, 0x83, 0x0F); }
-void jae(const char *label, LabelType type = T_AUTO) { jae(std::string(label), type); }
-void jae(const void *addr) { opJmpAbs(addr, T_NEAR, 0x73, 0x83, 0x0F); }
-void jae(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x73, 0x83, 0x0F); }
-void jb(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x72, 0x82, 0x0F); }
-void jb(const char *label, LabelType type = T_AUTO) { jb(std::string(label), type); }
-void jb(const void *addr) { opJmpAbs(addr, T_NEAR, 0x72, 0x82, 0x0F); }
-void jb(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x72, 0x82, 0x0F); }
-void jbe(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x76, 0x86, 0x0F); }
-void jbe(const char *label, LabelType type = T_AUTO) { jbe(std::string(label), type); }
-void jbe(const void *addr) { opJmpAbs(addr, T_NEAR, 0x76, 0x86, 0x0F); }
-void jbe(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x76, 0x86, 0x0F); }
-void jc(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x72, 0x82, 0x0F); }
-void jc(const char *label, LabelType type = T_AUTO) { jc(std::string(label), type); }
-void jc(const void *addr) { opJmpAbs(addr, T_NEAR, 0x72, 0x82, 0x0F); }
-void jc(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x72, 0x82, 0x0F); }
-void je(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x74, 0x84, 0x0F); }
-void je(const char *label, LabelType type = T_AUTO) { je(std::string(label), type); }
-void je(const void *addr) { opJmpAbs(addr, T_NEAR, 0x74, 0x84, 0x0F); }
-void je(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x74, 0x84, 0x0F); }
-void jg(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7F, 0x8F, 0x0F); }
-void jg(const char *label, LabelType type = T_AUTO) { jg(std::string(label), type); }
-void jg(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7F, 0x8F, 0x0F); }
-void jg(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7F, 0x8F, 0x0F); }
-void jge(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7D, 0x8D, 0x0F); }
-void jge(const char *label, LabelType type = T_AUTO) { jge(std::string(label), type); }
-void jge(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7D, 0x8D, 0x0F); }
-void jge(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7D, 0x8D, 0x0F); }
-void jl(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7C, 0x8C, 0x0F); }
-void jl(const char *label, LabelType type = T_AUTO) { jl(std::string(label), type); }
-void jl(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7C, 0x8C, 0x0F); }
-void jl(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7C, 0x8C, 0x0F); }
-void jle(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7E, 0x8E, 0x0F); }
-void jle(const char *label, LabelType type = T_AUTO) { jle(std::string(label), type); }
-void jle(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7E, 0x8E, 0x0F); }
-void jle(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7E, 0x8E, 0x0F); }
-void jna(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x76, 0x86, 0x0F); }
-void jna(const char *label, LabelType type = T_AUTO) { jna(std::string(label), type); }
-void jna(const void *addr) { opJmpAbs(addr, T_NEAR, 0x76, 0x86, 0x0F); }
-void jna(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x76, 0x86, 0x0F); }
-void jnae(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x72, 0x82, 0x0F); }
-void jnae(const char *label, LabelType type = T_AUTO) { jnae(std::string(label), type); }
-void jnae(const void *addr) { opJmpAbs(addr, T_NEAR, 0x72, 0x82, 0x0F); }
-void jnae(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x72, 0x82, 0x0F); }
-void jnb(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x73, 0x83, 0x0F); }
-void jnb(const char *label, LabelType type = T_AUTO) { jnb(std::string(label), type); }
-void jnb(const void *addr) { opJmpAbs(addr, T_NEAR, 0x73, 0x83, 0x0F); }
-void jnb(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x73, 0x83, 0x0F); }
-void jnbe(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x77, 0x87, 0x0F); }
-void jnbe(const char *label, LabelType type = T_AUTO) { jnbe(std::string(label), type); }
-void jnbe(const void *addr) { opJmpAbs(addr, T_NEAR, 0x77, 0x87, 0x0F); }
-void jnbe(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x77, 0x87, 0x0F); }
-void jnc(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x73, 0x83, 0x0F); }
-void jnc(const char *label, LabelType type = T_AUTO) { jnc(std::string(label), type); }
-void jnc(const void *addr) { opJmpAbs(addr, T_NEAR, 0x73, 0x83, 0x0F); }
-void jnc(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x73, 0x83, 0x0F); }
-void jne(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x75, 0x85, 0x0F); }
-void jne(const char *label, LabelType type = T_AUTO) { jne(std::string(label), type); }
-void jne(const void *addr) { opJmpAbs(addr, T_NEAR, 0x75, 0x85, 0x0F); }
-void jne(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x75, 0x85, 0x0F); }
-void jng(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7E, 0x8E, 0x0F); }
-void jng(const char *label, LabelType type = T_AUTO) { jng(std::string(label), type); }
-void jng(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7E, 0x8E, 0x0F); }
-void jng(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7E, 0x8E, 0x0F); }
-void jnge(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7C, 0x8C, 0x0F); }
-void jnge(const char *label, LabelType type = T_AUTO) { jnge(std::string(label), type); }
-void jnge(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7C, 0x8C, 0x0F); }
-void jnge(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7C, 0x8C, 0x0F); }
-void jnl(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7D, 0x8D, 0x0F); }
-void jnl(const char *label, LabelType type = T_AUTO) { jnl(std::string(label), type); }
-void jnl(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7D, 0x8D, 0x0F); }
-void jnl(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7D, 0x8D, 0x0F); }
-void jnle(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7F, 0x8F, 0x0F); }
-void jnle(const char *label, LabelType type = T_AUTO) { jnle(std::string(label), type); }
-void jnle(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7F, 0x8F, 0x0F); }
-void jnle(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7F, 0x8F, 0x0F); }
-void jno(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x71, 0x81, 0x0F); }
-void jno(const char *label, LabelType type = T_AUTO) { jno(std::string(label), type); }
-void jno(const void *addr) { opJmpAbs(addr, T_NEAR, 0x71, 0x81, 0x0F); }
-void jno(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x71, 0x81, 0x0F); }
-void jnp(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7B, 0x8B, 0x0F); }
-void jnp(const char *label, LabelType type = T_AUTO) { jnp(std::string(label), type); }
-void jnp(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7B, 0x8B, 0x0F); }
-void jnp(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7B, 0x8B, 0x0F); }
-void jns(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x79, 0x89, 0x0F); }
-void jns(const char *label, LabelType type = T_AUTO) { jns(std::string(label), type); }
-void jns(const void *addr) { opJmpAbs(addr, T_NEAR, 0x79, 0x89, 0x0F); }
-void jns(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x79, 0x89, 0x0F); }
-void jnz(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x75, 0x85, 0x0F); }
-void jnz(const char *label, LabelType type = T_AUTO) { jnz(std::string(label), type); }
-void jnz(const void *addr) { opJmpAbs(addr, T_NEAR, 0x75, 0x85, 0x0F); }
-void jnz(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x75, 0x85, 0x0F); }
-void jo(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x70, 0x80, 0x0F); }
-void jo(const char *label, LabelType type = T_AUTO) { jo(std::string(label), type); }
-void jo(const void *addr) { opJmpAbs(addr, T_NEAR, 0x70, 0x80, 0x0F); }
-void jo(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x70, 0x80, 0x0F); }
-void jp(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7A, 0x8A, 0x0F); }
-void jp(const char *label, LabelType type = T_AUTO) { jp(std::string(label), type); }
-void jp(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7A, 0x8A, 0x0F); }
-void jp(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7A, 0x8A, 0x0F); }
-void jpe(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7A, 0x8A, 0x0F); }
-void jpe(const char *label, LabelType type = T_AUTO) { jpe(std::string(label), type); }
-void jpe(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7A, 0x8A, 0x0F); }
-void jpe(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7A, 0x8A, 0x0F); }
-void jpo(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7B, 0x8B, 0x0F); }
-void jpo(const char *label, LabelType type = T_AUTO) { jpo(std::string(label), type); }
-void jpo(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7B, 0x8B, 0x0F); }
-void jpo(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7B, 0x8B, 0x0F); }
-void js(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x78, 0x88, 0x0F); }
-void js(const char *label, LabelType type = T_AUTO) { js(std::string(label), type); }
-void js(const void *addr) { opJmpAbs(addr, T_NEAR, 0x78, 0x88, 0x0F); }
-void js(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x78, 0x88, 0x0F); }
-void jz(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x74, 0x84, 0x0F); }
-void jz(const char *label, LabelType type = T_AUTO) { jz(std::string(label), type); }
-void jz(const void *addr) { opJmpAbs(addr, T_NEAR, 0x74, 0x84, 0x0F); }
-void jz(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x74, 0x84, 0x0F); }
+void insertps(const Xmm& xmm, const Operand& op, uint8_t imm) { opGen(xmm, op, 0x21, 0x66, isXMM_XMMorMEM, imm, 0x3A); }
+void int3() { db(0xCC); }
+void int_(uint8_t x) { db(0xCD); db(x); }
+void ja(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x77, 0x87, 0x0F); }//-V524
+void ja(const char *label, LabelType type = T_AUTO) { ja(std::string(label), type); }//-V524
+void ja(const void *addr) { opJmpAbs(addr, T_NEAR, 0x77, 0x87, 0x0F); }//-V524
+void ja(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x77, 0x87, 0x0F); }//-V524
+void jae(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x73, 0x83, 0x0F); }//-V524
+void jae(const char *label, LabelType type = T_AUTO) { jae(std::string(label), type); }//-V524
+void jae(const void *addr) { opJmpAbs(addr, T_NEAR, 0x73, 0x83, 0x0F); }//-V524
+void jae(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x73, 0x83, 0x0F); }//-V524
+void jb(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x72, 0x82, 0x0F); }//-V524
+void jb(const char *label, LabelType type = T_AUTO) { jb(std::string(label), type); }//-V524
+void jb(const void *addr) { opJmpAbs(addr, T_NEAR, 0x72, 0x82, 0x0F); }//-V524
+void jb(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x72, 0x82, 0x0F); }//-V524
+void jbe(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x76, 0x86, 0x0F); }//-V524
+void jbe(const char *label, LabelType type = T_AUTO) { jbe(std::string(label), type); }//-V524
+void jbe(const void *addr) { opJmpAbs(addr, T_NEAR, 0x76, 0x86, 0x0F); }//-V524
+void jbe(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x76, 0x86, 0x0F); }//-V524
+void jc(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x72, 0x82, 0x0F); }//-V524
+void jc(const char *label, LabelType type = T_AUTO) { jc(std::string(label), type); }//-V524
+void jc(const void *addr) { opJmpAbs(addr, T_NEAR, 0x72, 0x82, 0x0F); }//-V524
+void jc(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x72, 0x82, 0x0F); }//-V524
+void je(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x74, 0x84, 0x0F); }//-V524
+void je(const char *label, LabelType type = T_AUTO) { je(std::string(label), type); }//-V524
+void je(const void *addr) { opJmpAbs(addr, T_NEAR, 0x74, 0x84, 0x0F); }//-V524
+void je(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x74, 0x84, 0x0F); }//-V524
+void jg(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7F, 0x8F, 0x0F); }//-V524
+void jg(const char *label, LabelType type = T_AUTO) { jg(std::string(label), type); }//-V524
+void jg(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7F, 0x8F, 0x0F); }//-V524
+void jg(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7F, 0x8F, 0x0F); }//-V524
+void jge(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7D, 0x8D, 0x0F); }//-V524
+void jge(const char *label, LabelType type = T_AUTO) { jge(std::string(label), type); }//-V524
+void jge(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7D, 0x8D, 0x0F); }//-V524
+void jge(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7D, 0x8D, 0x0F); }//-V524
+void jl(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7C, 0x8C, 0x0F); }//-V524
+void jl(const char *label, LabelType type = T_AUTO) { jl(std::string(label), type); }//-V524
+void jl(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7C, 0x8C, 0x0F); }//-V524
+void jl(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7C, 0x8C, 0x0F); }//-V524
+void jle(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7E, 0x8E, 0x0F); }//-V524
+void jle(const char *label, LabelType type = T_AUTO) { jle(std::string(label), type); }//-V524
+void jle(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7E, 0x8E, 0x0F); }//-V524
+void jle(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7E, 0x8E, 0x0F); }//-V524
+void jna(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x76, 0x86, 0x0F); }//-V524
+void jna(const char *label, LabelType type = T_AUTO) { jna(std::string(label), type); }//-V524
+void jna(const void *addr) { opJmpAbs(addr, T_NEAR, 0x76, 0x86, 0x0F); }//-V524
+void jna(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x76, 0x86, 0x0F); }//-V524
+void jnae(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x72, 0x82, 0x0F); }//-V524
+void jnae(const char *label, LabelType type = T_AUTO) { jnae(std::string(label), type); }//-V524
+void jnae(const void *addr) { opJmpAbs(addr, T_NEAR, 0x72, 0x82, 0x0F); }//-V524
+void jnae(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x72, 0x82, 0x0F); }//-V524
+void jnb(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x73, 0x83, 0x0F); }//-V524
+void jnb(const char *label, LabelType type = T_AUTO) { jnb(std::string(label), type); }//-V524
+void jnb(const void *addr) { opJmpAbs(addr, T_NEAR, 0x73, 0x83, 0x0F); }//-V524
+void jnb(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x73, 0x83, 0x0F); }//-V524
+void jnbe(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x77, 0x87, 0x0F); }//-V524
+void jnbe(const char *label, LabelType type = T_AUTO) { jnbe(std::string(label), type); }//-V524
+void jnbe(const void *addr) { opJmpAbs(addr, T_NEAR, 0x77, 0x87, 0x0F); }//-V524
+void jnbe(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x77, 0x87, 0x0F); }//-V524
+void jnc(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x73, 0x83, 0x0F); }//-V524
+void jnc(const char *label, LabelType type = T_AUTO) { jnc(std::string(label), type); }//-V524
+void jnc(const void *addr) { opJmpAbs(addr, T_NEAR, 0x73, 0x83, 0x0F); }//-V524
+void jnc(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x73, 0x83, 0x0F); }//-V524
+void jne(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x75, 0x85, 0x0F); }//-V524
+void jne(const char *label, LabelType type = T_AUTO) { jne(std::string(label), type); }//-V524
+void jne(const void *addr) { opJmpAbs(addr, T_NEAR, 0x75, 0x85, 0x0F); }//-V524
+void jne(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x75, 0x85, 0x0F); }//-V524
+void jng(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7E, 0x8E, 0x0F); }//-V524
+void jng(const char *label, LabelType type = T_AUTO) { jng(std::string(label), type); }//-V524
+void jng(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7E, 0x8E, 0x0F); }//-V524
+void jng(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7E, 0x8E, 0x0F); }//-V524
+void jnge(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7C, 0x8C, 0x0F); }//-V524
+void jnge(const char *label, LabelType type = T_AUTO) { jnge(std::string(label), type); }//-V524
+void jnge(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7C, 0x8C, 0x0F); }//-V524
+void jnge(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7C, 0x8C, 0x0F); }//-V524
+void jnl(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7D, 0x8D, 0x0F); }//-V524
+void jnl(const char *label, LabelType type = T_AUTO) { jnl(std::string(label), type); }//-V524
+void jnl(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7D, 0x8D, 0x0F); }//-V524
+void jnl(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7D, 0x8D, 0x0F); }//-V524
+void jnle(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7F, 0x8F, 0x0F); }//-V524
+void jnle(const char *label, LabelType type = T_AUTO) { jnle(std::string(label), type); }//-V524
+void jnle(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7F, 0x8F, 0x0F); }//-V524
+void jnle(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7F, 0x8F, 0x0F); }//-V524
+void jno(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x71, 0x81, 0x0F); }//-V524
+void jno(const char *label, LabelType type = T_AUTO) { jno(std::string(label), type); }//-V524
+void jno(const void *addr) { opJmpAbs(addr, T_NEAR, 0x71, 0x81, 0x0F); }//-V524
+void jno(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x71, 0x81, 0x0F); }//-V524
+void jnp(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7B, 0x8B, 0x0F); }//-V524
+void jnp(const char *label, LabelType type = T_AUTO) { jnp(std::string(label), type); }//-V524
+void jnp(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7B, 0x8B, 0x0F); }//-V524
+void jnp(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7B, 0x8B, 0x0F); }//-V524
+void jns(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x79, 0x89, 0x0F); }//-V524
+void jns(const char *label, LabelType type = T_AUTO) { jns(std::string(label), type); }//-V524
+void jns(const void *addr) { opJmpAbs(addr, T_NEAR, 0x79, 0x89, 0x0F); }//-V524
+void jns(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x79, 0x89, 0x0F); }//-V524
+void jnz(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x75, 0x85, 0x0F); }//-V524
+void jnz(const char *label, LabelType type = T_AUTO) { jnz(std::string(label), type); }//-V524
+void jnz(const void *addr) { opJmpAbs(addr, T_NEAR, 0x75, 0x85, 0x0F); }//-V524
+void jnz(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x75, 0x85, 0x0F); }//-V524
+void jo(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x70, 0x80, 0x0F); }//-V524
+void jo(const char *label, LabelType type = T_AUTO) { jo(std::string(label), type); }//-V524
+void jo(const void *addr) { opJmpAbs(addr, T_NEAR, 0x70, 0x80, 0x0F); }//-V524
+void jo(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x70, 0x80, 0x0F); }//-V524
+void jp(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7A, 0x8A, 0x0F); }//-V524
+void jp(const char *label, LabelType type = T_AUTO) { jp(std::string(label), type); }//-V524
+void jp(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7A, 0x8A, 0x0F); }//-V524
+void jp(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7A, 0x8A, 0x0F); }//-V524
+void jpe(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7A, 0x8A, 0x0F); }//-V524
+void jpe(const char *label, LabelType type = T_AUTO) { jpe(std::string(label), type); }//-V524
+void jpe(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7A, 0x8A, 0x0F); }//-V524
+void jpe(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7A, 0x8A, 0x0F); }//-V524
+void jpo(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x7B, 0x8B, 0x0F); }//-V524
+void jpo(const char *label, LabelType type = T_AUTO) { jpo(std::string(label), type); }//-V524
+void jpo(const void *addr) { opJmpAbs(addr, T_NEAR, 0x7B, 0x8B, 0x0F); }//-V524
+void jpo(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x7B, 0x8B, 0x0F); }//-V524
+void js(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x78, 0x88, 0x0F); }//-V524
+void js(const char *label, LabelType type = T_AUTO) { js(std::string(label), type); }//-V524
+void js(const void *addr) { opJmpAbs(addr, T_NEAR, 0x78, 0x88, 0x0F); }//-V524
+void js(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x78, 0x88, 0x0F); }//-V524
+void jz(const Label& label, LabelType type = T_AUTO) { opJmp(label, type, 0x74, 0x84, 0x0F); }//-V524
+void jz(const char *label, LabelType type = T_AUTO) { jz(std::string(label), type); }//-V524
+void jz(const void *addr) { opJmpAbs(addr, T_NEAR, 0x74, 0x84, 0x0F); }//-V524
+void jz(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0x74, 0x84, 0x0F); }//-V524
 void lahf() { db(0x9F); }
 void lddqu(const Xmm& xmm, const Address& addr) { db(0xF2); opModM(addr, xmm, 0x0F, 0xF0); }
 void ldmxcsr(const Address& addr) { opModM(addr, Reg32(2), 0x0F, 0xAE); }
-void lea(const Reg& reg, const Address& addr) { if (!reg.isBit(16 | i32e)) throw Error(ERR_BAD_SIZE_OF_REGISTER); opModM(addr, reg, 0x8D); }
+void lea(const Reg& reg, const Address& addr) { if (!reg.isBit(16 | i32e)) XBYAK_THROW(ERR_BAD_SIZE_OF_REGISTER) opModM(addr, reg, 0x8D); }
+void leave() { db(0xC9); }
 void lfence() { db(0x0F); db(0xAE); db(0xE8); }
+void lfs(const Reg& reg, const Address& addr) { opLoadSeg(addr, reg, 0x0F, 0xB4); }
+void lgs(const Reg& reg, const Address& addr) { opLoadSeg(addr, reg, 0x0F, 0xB5); }
 void lock() { db(0xF0); }
+void lodsb() { db(0xAC); }
+void lodsd() { db(0xAD); }
+void lodsw() { db(0x66); db(0xAD); }
+void loop(const Label& label) { opJmp(label, T_SHORT, 0xE2, 0, 0); }
+void loop(const char *label) { loop(std::string(label)); }
+void loop(std::string label) { opJmp(label, T_SHORT, 0xE2, 0, 0); }
+void loope(const Label& label) { opJmp(label, T_SHORT, 0xE1, 0, 0); }
+void loope(const char *label) { loope(std::string(label)); }
+void loope(std::string label) { opJmp(label, T_SHORT, 0xE1, 0, 0); }
+void loopne(const Label& label) { opJmp(label, T_SHORT, 0xE0, 0, 0); }
+void loopne(const char *label) { loopne(std::string(label)); }
+void loopne(std::string label) { opJmp(label, T_SHORT, 0xE0, 0, 0); }
+void lss(const Reg& reg, const Address& addr) { opLoadSeg(addr, reg, 0x0F, 0xB2); }
 void lzcnt(const Reg&reg, const Operand& op) { opSp1(reg, op, 0xF3, 0x0F, 0xBD); }
 void maskmovdqu(const Xmm& reg1, const Xmm& reg2) { db(0x66);  opModR(reg1, reg2, 0x0F, 0xF7); }
-void maskmovq(const Mmx& reg1, const Mmx& reg2) { if (!reg1.isMMX() || !reg2.isMMX()) throw Error(ERR_BAD_COMBINATION); opModR(reg1, reg2, 0x0F, 0xF7); }
+void maskmovq(const Mmx& reg1, const Mmx& reg2) { if (!reg1.isMMX() || !reg2.isMMX()) XBYAK_THROW(ERR_BAD_COMBINATION) opModR(reg1, reg2, 0x0F, 0xF7); }
 void maxpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x5F, 0x66, isXMM_XMMorMEM); }
 void maxps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x5F, 0x100, isXMM_XMMorMEM); }
 void maxsd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x5F, 0xF2, isXMM_XMMorMEM); }
@@ -432,6 +495,7 @@ void minps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x5D, 0x100, isXM
 void minsd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x5D, 0xF2, isXMM_XMMorMEM); }
 void minss(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x5D, 0xF3, isXMM_XMMorMEM); }
 void monitor() { db(0x0F); db(0x01); db(0xC8); }
+void monitorx() { db(0x0F); db(0x01); db(0xFA); }
 void movapd(const Address& addr, const Xmm& xmm) { db(0x66); opModM(addr, xmm, 0x0F, 0x29); }
 void movapd(const Xmm& xmm, const Operand& op) { opMMX(xmm, op, 0x28, 0x66); }
 void movaps(const Address& addr, const Xmm& xmm) { opModM(addr, xmm, 0x0F, 0x29); }
@@ -442,7 +506,9 @@ void movd(const Address& addr, const Mmx& mmx) { if (mmx.isXMM()) db(0x66); opMo
 void movd(const Mmx& mmx, const Address& addr) { if (mmx.isXMM()) db(0x66); opModM(addr, mmx, 0x0F, 0x6E); }
 void movd(const Mmx& mmx, const Reg32& reg) { if (mmx.isXMM()) db(0x66); opModR(mmx, reg, 0x0F, 0x6E); }
 void movd(const Reg32& reg, const Mmx& mmx) { if (mmx.isXMM()) db(0x66); opModR(mmx, reg, 0x0F, 0x7E); }
-void movddup(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x12, 0xF2, isXMM_XMMorMEM); }
+void movddup(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x12, 0xF2, isXMM_XMMorMEM, NONE, NONE); }
+void movdir64b(const Reg& reg, const Address& addr) { db(0x66); opModM(addr, reg.cvt32(), 0x0F, 0x38, 0xF8); }
+void movdiri(const Address& addr, const Reg32e& reg) { opModM(addr, reg, 0x0F, 0x38, 0xF9); }
 void movdq2q(const Mmx& mmx, const Xmm& xmm) { db(0xF2); opModR(mmx, xmm, 0x0F, 0xD6); }
 void movdqa(const Address& addr, const Xmm& xmm) { db(0x66); opModM(addr, xmm, 0x0F, 0x7F); }
 void movdqa(const Xmm& xmm, const Operand& op) { opMMX(xmm, op, 0x6F, 0x66); }
@@ -461,23 +527,26 @@ void movntdqa(const Xmm& xmm, const Address& addr) { db(0x66); opModM(addr, xmm,
 void movnti(const Address& addr, const Reg32e& reg) { opModM(addr, reg, 0x0F, 0xC3); }
 void movntpd(const Address& addr, const Xmm& reg) { opModM(addr, Reg16(reg.getIdx()), 0x0F, 0x2B); }
 void movntps(const Address& addr, const Xmm& xmm) { opModM(addr, Mmx(xmm.getIdx()), 0x0F, 0x2B); }
-void movntq(const Address& addr, const Mmx& mmx) { if (!mmx.isMMX()) throw Error(ERR_BAD_COMBINATION); opModM(addr, mmx, 0x0F, 0xE7); }
+void movntq(const Address& addr, const Mmx& mmx) { if (!mmx.isMMX()) XBYAK_THROW(ERR_BAD_COMBINATION) opModM(addr, mmx, 0x0F, 0xE7); }
 void movq(const Address& addr, const Mmx& mmx) { if (mmx.isXMM()) db(0x66); opModM(addr, mmx, 0x0F, mmx.isXMM() ? 0xD6 : 0x7F); }
 void movq(const Mmx& mmx, const Operand& op) { if (mmx.isXMM()) db(0xF3); opModRM(mmx, op, (mmx.getKind() == op.getKind()), op.isMEM(), 0x0F, mmx.isXMM() ? 0x7E : 0x6F); }
 void movq2dq(const Xmm& xmm, const Mmx& mmx) { db(0xF3); opModR(xmm, mmx, 0x0F, 0xD6); }
+void movsb() { db(0xA4); }
+void movsd() { db(0xA5); }
 void movsd(const Address& addr, const Xmm& xmm) { db(0xF2); opModM(addr, xmm, 0x0F, 0x11); }
 void movsd(const Xmm& xmm, const Operand& op) { opMMX(xmm, op, 0x10, 0xF2); }
-void movshdup(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x16, 0xF3, isXMM_XMMorMEM); }
-void movsldup(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x12, 0xF3, isXMM_XMMorMEM); }
+void movshdup(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x16, 0xF3, isXMM_XMMorMEM, NONE, NONE); }
+void movsldup(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x12, 0xF3, isXMM_XMMorMEM, NONE, NONE); }
 void movss(const Address& addr, const Xmm& xmm) { db(0xF3); opModM(addr, xmm, 0x0F, 0x11); }
 void movss(const Xmm& xmm, const Operand& op) { opMMX(xmm, op, 0x10, 0xF3); }
+void movsw() { db(0x66); db(0xA5); }
 void movsx(const Reg& reg, const Operand& op) { opMovxx(reg, op, 0xBE); }
 void movupd(const Address& addr, const Xmm& xmm) { db(0x66); opModM(addr, xmm, 0x0F, 0x11); }
 void movupd(const Xmm& xmm, const Operand& op) { opMMX(xmm, op, 0x10, 0x66); }
 void movups(const Address& addr, const Xmm& xmm) { opModM(addr, xmm, 0x0F, 0x11); }
 void movups(const Xmm& xmm, const Operand& op) { opMMX(xmm, op, 0x10, 0x100); }
 void movzx(const Reg& reg, const Operand& op) { opMovxx(reg, op, 0xB6); }
-void mpsadbw(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x42, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
+void mpsadbw(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x42, 0x66, isXMM_XMMorMEM, static_cast<uint8_t>(imm), 0x3A); }
 void mul(const Operand& op) { opR_ModM(op, 0, 4, 0xF6); }
 void mulpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x59, 0x66, isXMM_XMMorMEM); }
 void mulps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x59, 0x100, isXMM_XMMorMEM); }
@@ -485,13 +554,18 @@ void mulsd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x59, 0xF2, isXMM
 void mulss(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x59, 0xF3, isXMM_XMMorMEM); }
 void mulx(const Reg32e& r1, const Reg32e& r2, const Operand& op) { opGpr(r1, r2, op, T_F2 | T_0F38, 0xf6, true); }
 void mwait() { db(0x0F); db(0x01); db(0xC9); }
+void mwaitx() { db(0x0F); db(0x01); db(0xFB); }
 void neg(const Operand& op) { opR_ModM(op, 0, 3, 0xF6); }
-void nop() { db(0x90); }
 void not_(const Operand& op) { opR_ModM(op, 0, 2, 0xF6); }
-void or_(const Operand& op, uint32 imm) { opRM_I(op, imm, 0x08, 1); }
+void or_(const Operand& op, uint32_t imm) { opRM_I(op, imm, 0x08, 1); }
 void or_(const Operand& op1, const Operand& op2) { opRM_RM(op1, op2, 0x08); }
 void orpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x56, 0x66, isXMM_XMMorMEM); }
 void orps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x56, 0x100, isXMM_XMMorMEM); }
+void out_(const Reg& d, const Reg& a) { opInOut(a, d, 0xEE); }
+void out_(uint8_t v, const Reg& a) { opInOut(a, 0xE6, v); }
+void outsb() { db(0x6E); }
+void outsd() { db(0x6F); }
+void outsw() { db(0x66); db(0x6F); }
 void pabsb(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x1C, 0x66, NONE, 0x38); }
 void pabsd(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x1E, 0x66, NONE, 0x38); }
 void pabsw(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x1D, 0x66, NONE, 0x38); }
@@ -507,36 +581,36 @@ void paddsw(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0xED); }
 void paddusb(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0xDC); }
 void paddusw(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0xDD); }
 void paddw(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0xFD); }
-void palignr(const Mmx& mmx, const Operand& op, int imm) { opMMX(mmx, op, 0x0f, 0x66, static_cast<uint8>(imm), 0x3a); }
+void palignr(const Mmx& mmx, const Operand& op, int imm) { opMMX(mmx, op, 0x0f, 0x66, static_cast<uint8_t>(imm), 0x3a); }
 void pand(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0xDB); }
 void pandn(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0xDF); }
 void pause() { db(0xF3); db(0x90); }
 void pavgb(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0xE0); }
 void pavgw(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0xE3); }
 void pblendvb(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x10, 0x66, isXMM_XMMorMEM, NONE, 0x38); }
-void pblendw(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x0E, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
-void pclmulhqhdq(const Xmm& xmm, const Operand& op) { pclmulqdq(xmm, op, 0x11); }
+void pblendw(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x0E, 0x66, isXMM_XMMorMEM, static_cast<uint8_t>(imm), 0x3A); }
+void pclmulhqhqdq(const Xmm& xmm, const Operand& op) { pclmulqdq(xmm, op, 0x11); }
 void pclmulhqlqdq(const Xmm& xmm, const Operand& op) { pclmulqdq(xmm, op, 0x01); }
-void pclmullqhdq(const Xmm& xmm, const Operand& op) { pclmulqdq(xmm, op, 0x10); }
+void pclmullqhqdq(const Xmm& xmm, const Operand& op) { pclmulqdq(xmm, op, 0x10); }
 void pclmullqlqdq(const Xmm& xmm, const Operand& op) { pclmulqdq(xmm, op, 0x00); }
-void pclmulqdq(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x44, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
+void pclmulqdq(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x44, 0x66, isXMM_XMMorMEM, static_cast<uint8_t>(imm), 0x3A); }
 void pcmpeqb(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x74); }
 void pcmpeqd(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x76); }
 void pcmpeqq(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x29, 0x66, isXMM_XMMorMEM, NONE, 0x38); }
 void pcmpeqw(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x75); }
-void pcmpestri(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x61, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
-void pcmpestrm(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x60, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
+void pcmpestri(const Xmm& xmm, const Operand& op, uint8_t imm) { opGen(xmm, op, 0x61, 0x66, isXMM_XMMorMEM, imm, 0x3A); }
+void pcmpestrm(const Xmm& xmm, const Operand& op, uint8_t imm) { opGen(xmm, op, 0x60, 0x66, isXMM_XMMorMEM, imm, 0x3A); }
 void pcmpgtb(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x64); }
 void pcmpgtd(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x66); }
 void pcmpgtq(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x37, 0x66, isXMM_XMMorMEM, NONE, 0x38); }
 void pcmpgtw(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x65); }
-void pcmpistri(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x63, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
-void pcmpistrm(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x62, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
+void pcmpistri(const Xmm& xmm, const Operand& op, uint8_t imm) { opGen(xmm, op, 0x63, 0x66, isXMM_XMMorMEM, imm, 0x3A); }
+void pcmpistrm(const Xmm& xmm, const Operand& op, uint8_t imm) { opGen(xmm, op, 0x62, 0x66, isXMM_XMMorMEM, imm, 0x3A); }
 void pdep(const Reg32e& r1, const Reg32e& r2, const Operand& op) { opGpr(r1, r2, op, T_F2 | T_0F38, 0xf5, true); }
 void pext(const Reg32e& r1, const Reg32e& r2, const Operand& op) { opGpr(r1, r2, op, T_F3 | T_0F38, 0xf5, true); }
-void pextrb(const Operand& op, const Xmm& xmm, uint8 imm) { opExt(op, xmm, 0x14, imm); }
-void pextrd(const Operand& op, const Xmm& xmm, uint8 imm) { opExt(op, xmm, 0x16, imm); }
-void pextrw(const Operand& op, const Mmx& xmm, uint8 imm) { opExt(op, xmm, 0x15, imm, true); }
+void pextrb(const Operand& op, const Xmm& xmm, uint8_t imm) { opExt(op, xmm, 0x14, imm); }
+void pextrd(const Operand& op, const Xmm& xmm, uint8_t imm) { opExt(op, xmm, 0x16, imm); }
+void pextrw(const Operand& op, const Mmx& xmm, uint8_t imm) { opExt(op, xmm, 0x15, imm, true); }
 void phaddd(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x02, 0x66, NONE, 0x38); }
 void phaddsw(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x03, 0x66, NONE, 0x38); }
 void phaddw(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x01, 0x66, NONE, 0x38); }
@@ -544,9 +618,9 @@ void phminposuw(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x41, 0x66,
 void phsubd(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x06, 0x66, NONE, 0x38); }
 void phsubsw(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x07, 0x66, NONE, 0x38); }
 void phsubw(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x05, 0x66, NONE, 0x38); }
-void pinsrb(const Xmm& xmm, const Operand& op, uint8 imm) { opGen(xmm, op, 0x20, 0x66, isXMM_REG32orMEM, imm, 0x3A); }
-void pinsrd(const Xmm& xmm, const Operand& op, uint8 imm) { opGen(xmm, op, 0x22, 0x66, isXMM_REG32orMEM, imm, 0x3A); }
-void pinsrw(const Mmx& mmx, const Operand& op, int imm) { if (!op.isREG(32) && !op.isMEM()) throw Error(ERR_BAD_COMBINATION); opGen(mmx, op, 0xC4, mmx.isXMM() ? 0x66 : NONE, 0, imm); }
+void pinsrb(const Xmm& xmm, const Operand& op, uint8_t imm) { opGen(xmm, op, 0x20, 0x66, isXMM_REG32orMEM, imm, 0x3A); }
+void pinsrd(const Xmm& xmm, const Operand& op, uint8_t imm) { opGen(xmm, op, 0x22, 0x66, isXMM_REG32orMEM, imm, 0x3A); }
+void pinsrw(const Mmx& mmx, const Operand& op, int imm) { if (!op.isREG(32) && !op.isMEM()) XBYAK_THROW(ERR_BAD_COMBINATION) opGen(mmx, op, 0xC4, mmx.isXMM() ? 0x66 : NONE, 0, imm); }
 void pmaddubsw(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x04, 0x66, NONE, 0x38); }
 void pmaddwd(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0xF5); }
 void pmaxsb(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x3C, 0x66, isXMM_XMMorMEM, NONE, 0x38); }
@@ -584,17 +658,20 @@ void pmuludq(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0xF4); }
 void popcnt(const Reg&reg, const Operand& op) { opSp1(reg, op, 0xF3, 0x0F, 0xB8); }
 void popf() { db(0x9D); }
 void por(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0xEB); }
+void prefetchit0(const Address& addr) { opModM(addr, Reg32(7), 0x0F, 0x18); }
+void prefetchit1(const Address& addr) { opModM(addr, Reg32(6), 0x0F, 0x18); }
 void prefetchnta(const Address& addr) { opModM(addr, Reg32(0), 0x0F, 0x18); }
 void prefetcht0(const Address& addr) { opModM(addr, Reg32(1), 0x0F, 0x18); }
 void prefetcht1(const Address& addr) { opModM(addr, Reg32(2), 0x0F, 0x18); }
 void prefetcht2(const Address& addr) { opModM(addr, Reg32(3), 0x0F, 0x18); }
+void prefetchw(const Address& addr) { opModM(addr, Reg32(1), 0x0F, 0x0D); }
 void prefetchwt1(const Address& addr) { opModM(addr, Reg32(2), 0x0F, 0x0D); }
 void psadbw(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0xF6); }
 void pshufb(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x00, 0x66, NONE, 0x38); }
-void pshufd(const Mmx& mmx, const Operand& op, uint8 imm8) { opMMX(mmx, op, 0x70, 0x66, imm8); }
-void pshufhw(const Mmx& mmx, const Operand& op, uint8 imm8) { opMMX(mmx, op, 0x70, 0xF3, imm8); }
-void pshuflw(const Mmx& mmx, const Operand& op, uint8 imm8) { opMMX(mmx, op, 0x70, 0xF2, imm8); }
-void pshufw(const Mmx& mmx, const Operand& op, uint8 imm8) { opMMX(mmx, op, 0x70, 0x00, imm8); }
+void pshufd(const Mmx& mmx, const Operand& op, uint8_t imm8) { opMMX(mmx, op, 0x70, 0x66, imm8); }
+void pshufhw(const Mmx& mmx, const Operand& op, uint8_t imm8) { opMMX(mmx, op, 0x70, 0xF3, imm8); }
+void pshuflw(const Mmx& mmx, const Operand& op, uint8_t imm8) { opMMX(mmx, op, 0x70, 0xF2, imm8); }
+void pshufw(const Mmx& mmx, const Operand& op, uint8_t imm8) { opMMX(mmx, op, 0x70, 0x00, imm8); }
 void psignb(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x08, 0x66, NONE, 0x38); }
 void psignd(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x0A, 0x66, NONE, 0x38); }
 void psignw(const Mmx& mmx, const Operand& op) { opMMX(mmx, op, 0x09, 0x66, NONE, 0x38); }
@@ -643,20 +720,26 @@ void rcr(const Operand& op, const Reg8& _cl) { opShift(op, _cl, 3); }
 void rcr(const Operand& op, int imm) { opShift(op, imm, 3); }
 void rdmsr() { db(0x0F); db(0x32); }
 void rdpmc() { db(0x0F); db(0x33); }
-void rdrand(const Reg& r) { if (r.isBit(8)) throw Error(ERR_BAD_SIZE_OF_REGISTER); opModR(Reg(6, Operand::REG, r.getBit()), r, 0x0F, 0xC7); }
-void rdseed(const Reg& r) { if (r.isBit(8)) throw Error(ERR_BAD_SIZE_OF_REGISTER); opModR(Reg(7, Operand::REG, r.getBit()), r, 0x0F, 0xC7); }
+void rdrand(const Reg& r) { if (r.isBit(8)) XBYAK_THROW(ERR_BAD_SIZE_OF_REGISTER) opModR(Reg(6, Operand::REG, r.getBit()), r, 0x0F, 0xC7); }
+void rdseed(const Reg& r) { if (r.isBit(8)) XBYAK_THROW(ERR_BAD_SIZE_OF_REGISTER) opModR(Reg(7, Operand::REG, r.getBit()), r, 0x0F, 0xC7); }
 void rdtsc() { db(0x0F); db(0x31); }
 void rdtscp() { db(0x0F); db(0x01); db(0xF9); }
+void rep() { db(0xF3); }
+void repe() { db(0xF3); }
+void repne() { db(0xF2); }
+void repnz() { db(0xF2); }
+void repz() { db(0xF3); }
 void ret(int imm = 0) { if (imm) { db(0xC2); dw(imm); } else { db(0xC3); } }
+void retf(int imm = 0) { if (imm) { db(0xCA); dw(imm); } else { db(0xCB); } }
 void rol(const Operand& op, const Reg8& _cl) { opShift(op, _cl, 0); }
 void rol(const Operand& op, int imm) { opShift(op, imm, 0); }
 void ror(const Operand& op, const Reg8& _cl) { opShift(op, _cl, 1); }
 void ror(const Operand& op, int imm) { opShift(op, imm, 1); }
-void rorx(const Reg32e& r, const Operand& op, uint8 imm) { opGpr(r, op, Reg32e(0, r.getBit()), T_0F3A | T_F2, 0xF0, false, imm); }
-void roundpd(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x09, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
-void roundps(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x08, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
-void roundsd(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x0B, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
-void roundss(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x0A, 0x66, isXMM_XMMorMEM, static_cast<uint8>(imm), 0x3A); }
+void rorx(const Reg32e& r, const Operand& op, uint8_t imm) { opGpr(r, op, Reg32e(0, r.getBit()), T_0F3A | T_F2, 0xF0, false, imm); }
+void roundpd(const Xmm& xmm, const Operand& op, uint8_t imm) { opGen(xmm, op, 0x09, 0x66, isXMM_XMMorMEM, imm, 0x3A); }
+void roundps(const Xmm& xmm, const Operand& op, uint8_t imm) { opGen(xmm, op, 0x08, 0x66, isXMM_XMMorMEM, imm, 0x3A); }
+void roundsd(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x0B, 0x66, isXMM_XMMorMEM, static_cast<uint8_t>(imm), 0x3A); }
+void roundss(const Xmm& xmm, const Operand& op, int imm) { opGen(xmm, op, 0x0A, 0x66, isXMM_XMMorMEM, static_cast<uint8_t>(imm), 0x3A); }
 void rsqrtps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x52, 0x100, isXMM_XMMorMEM); }
 void rsqrtss(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x52, 0xF3, isXMM_XMMorMEM); }
 void sahf() { db(0x9E); }
@@ -665,51 +748,62 @@ void sal(const Operand& op, int imm) { opShift(op, imm, 4); }
 void sar(const Operand& op, const Reg8& _cl) { opShift(op, _cl, 7); }
 void sar(const Operand& op, int imm) { opShift(op, imm, 7); }
 void sarx(const Reg32e& r1, const Operand& op, const Reg32e& r2) { opGpr(r1, op, r2, T_F3 | T_0F38, 0xf7, false); }
-void sbb(const Operand& op, uint32 imm) { opRM_I(op, imm, 0x18, 3); }
+void sbb(const Operand& op, uint32_t imm) { opRM_I(op, imm, 0x18, 3); }
 void sbb(const Operand& op1, const Operand& op2) { opRM_RM(op1, op2, 0x18); }
-void seta(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 7); }
-void setae(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 3); }
-void setb(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 2); }
-void setbe(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 6); }
-void setc(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 2); }
-void sete(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 4); }
-void setg(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 15); }
-void setge(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 13); }
-void setl(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 12); }
-void setle(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 14); }
-void setna(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 6); }
-void setnae(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 2); }
-void setnb(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 3); }
-void setnbe(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 7); }
-void setnc(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 3); }
-void setne(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 5); }
-void setng(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 14); }
-void setnge(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 12); }
-void setnl(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 13); }
-void setnle(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 15); }
-void setno(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 1); }
-void setnp(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 11); }
-void setns(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 9); }
-void setnz(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 5); }
-void seto(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 0); }
-void setp(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 10); }
-void setpe(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 10); }
-void setpo(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 11); }
-void sets(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 8); }
-void setz(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 4); }
+void scasb() { db(0xAE); }
+void scasd() { db(0xAF); }
+void scasw() { db(0x66); db(0xAF); }
+void serialize() { db(0x0F); db(0x01); db(0xE8); }
+void seta(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 7); }//-V524
+void setae(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 3); }//-V524
+void setb(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 2); }//-V524
+void setbe(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 6); }//-V524
+void setc(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 2); }//-V524
+void sete(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 4); }//-V524
+void setg(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 15); }//-V524
+void setge(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 13); }//-V524
+void setl(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 12); }//-V524
+void setle(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 14); }//-V524
+void setna(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 6); }//-V524
+void setnae(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 2); }//-V524
+void setnb(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 3); }//-V524
+void setnbe(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 7); }//-V524
+void setnc(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 3); }//-V524
+void setne(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 5); }//-V524
+void setng(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 14); }//-V524
+void setnge(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 12); }//-V524
+void setnl(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 13); }//-V524
+void setnle(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 15); }//-V524
+void setno(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 1); }//-V524
+void setnp(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 11); }//-V524
+void setns(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 9); }//-V524
+void setnz(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 5); }//-V524
+void seto(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 0); }//-V524
+void setp(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 10); }//-V524
+void setpe(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 10); }//-V524
+void setpo(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 11); }//-V524
+void sets(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 8); }//-V524
+void setz(const Operand& op) { opR_ModM(op, 8, 0, 0x0F, 0x90 | 4); }//-V524
 void sfence() { db(0x0F); db(0xAE); db(0xF8); }
+void sha1msg1(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0xC9, NONE, isXMM_XMMorMEM, NONE, 0x38); }
+void sha1msg2(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0xCA, NONE, isXMM_XMMorMEM, NONE, 0x38); }
+void sha1nexte(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0xC8, NONE, isXMM_XMMorMEM, NONE, 0x38); }
+void sha1rnds4(const Xmm& xmm, const Operand& op, uint8_t imm) { opGen(xmm, op, 0xCC, NONE, isXMM_XMMorMEM, imm, 0x3A); }
+void sha256msg1(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0xCC, NONE, isXMM_XMMorMEM, NONE, 0x38); }
+void sha256msg2(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0xCD, NONE, isXMM_XMMorMEM, NONE, 0x38); }
+void sha256rnds2(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0xCB, NONE, isXMM_XMMorMEM, NONE, 0x38); }
 void shl(const Operand& op, const Reg8& _cl) { opShift(op, _cl, 4); }
 void shl(const Operand& op, int imm) { opShift(op, imm, 4); }
 void shld(const Operand& op, const Reg& reg, const Reg8& _cl) { opShxd(op, reg, 0, 0xA4, &_cl); }
-void shld(const Operand& op, const Reg& reg, uint8 imm) { opShxd(op, reg, imm, 0xA4); }
+void shld(const Operand& op, const Reg& reg, uint8_t imm) { opShxd(op, reg, imm, 0xA4); }
 void shlx(const Reg32e& r1, const Operand& op, const Reg32e& r2) { opGpr(r1, op, r2, T_66 | T_0F38, 0xf7, false); }
 void shr(const Operand& op, const Reg8& _cl) { opShift(op, _cl, 5); }
 void shr(const Operand& op, int imm) { opShift(op, imm, 5); }
 void shrd(const Operand& op, const Reg& reg, const Reg8& _cl) { opShxd(op, reg, 0, 0xAC, &_cl); }
-void shrd(const Operand& op, const Reg& reg, uint8 imm) { opShxd(op, reg, imm, 0xAC); }
+void shrd(const Operand& op, const Reg& reg, uint8_t imm) { opShxd(op, reg, imm, 0xAC); }
 void shrx(const Reg32e& r1, const Operand& op, const Reg32e& r2) { opGpr(r1, op, r2, T_F2 | T_0F38, 0xf7, false); }
-void shufpd(const Xmm& xmm, const Operand& op, uint8 imm8) { opGen(xmm, op, 0xC6, 0x66, isXMM_XMMorMEM, imm8); }
-void shufps(const Xmm& xmm, const Operand& op, uint8 imm8) { opGen(xmm, op, 0xC6, 0x100, isXMM_XMMorMEM, imm8); }
+void shufpd(const Xmm& xmm, const Operand& op, uint8_t imm8) { opGen(xmm, op, 0xC6, 0x66, isXMM_XMMorMEM, imm8); }
+void shufps(const Xmm& xmm, const Operand& op, uint8_t imm8) { opGen(xmm, op, 0xC6, 0x100, isXMM_XMMorMEM, imm8); }
 void sqrtpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x51, 0x66, isXMM_XMMorMEM); }
 void sqrtps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x51, 0x100, isXMM_XMMorMEM); }
 void sqrtsd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x51, 0xF2, isXMM_XMMorMEM); }
@@ -719,44 +813,54 @@ void stc() { db(0xF9); }
 void std() { db(0xFD); }
 void sti() { db(0xFB); }
 void stmxcsr(const Address& addr) { opModM(addr, Reg32(3), 0x0F, 0xAE); }
-void sub(const Operand& op, uint32 imm) { opRM_I(op, imm, 0x28, 5); }
+void stosb() { db(0xAA); }
+void stosd() { db(0xAB); }
+void stosw() { db(0x66); db(0xAB); }
+void sub(const Operand& op, uint32_t imm) { opRM_I(op, imm, 0x28, 5); }
 void sub(const Operand& op1, const Operand& op2) { opRM_RM(op1, op2, 0x28); }
 void subpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x5C, 0x66, isXMM_XMMorMEM); }
 void subps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x5C, 0x100, isXMM_XMMorMEM); }
 void subsd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x5C, 0xF2, isXMM_XMMorMEM); }
 void subss(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x5C, 0xF3, isXMM_XMMorMEM); }
+void sysenter() { db(0x0F); db(0x34); }
+void sysexit() { db(0x0F); db(0x35); }
+void tpause(const Reg32& r) { int idx = r.getIdx(); if (idx > 7) XBYAK_THROW(ERR_BAD_PARAMETER) db(0x66); db(0x0F); db(0xAE); setModRM(3, 6, idx); }
 void tzcnt(const Reg&reg, const Operand& op) { opSp1(reg, op, 0xF3, 0x0F, 0xBC); }
 void ucomisd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x2E, 0x66, isXMM_XMMorMEM); }
 void ucomiss(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x2E, 0x100, isXMM_XMMorMEM); }
 void ud2() { db(0x0F); db(0x0B); }
+void umonitor(const Reg& r) { int idx = r.getIdx(); if (idx > 7) XBYAK_THROW(ERR_BAD_PARAMETER) int bit = r.getBit(); if (BIT != bit) { if ((BIT == 32 && bit == 16) || (BIT == 64 && bit == 32)) { db(0x67); } else { XBYAK_THROW(ERR_BAD_SIZE_OF_REGISTER) } } db(0xF3); db(0x0F); db(0xAE); setModRM(3, 6, idx); }
+void umwait(const Reg32& r) { int idx = r.getIdx(); if (idx > 7) XBYAK_THROW(ERR_BAD_PARAMETER) db(0xF2); db(0x0F); db(0xAE); setModRM(3, 6, idx); }
 void unpckhpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x15, 0x66, isXMM_XMMorMEM); }
 void unpckhps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x15, 0x100, isXMM_XMMorMEM); }
 void unpcklpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x14, 0x66, isXMM_XMMorMEM); }
 void unpcklps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x14, 0x100, isXMM_XMMorMEM); }
 void vaddpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER_Z | T_B64, 0x58); }
 void vaddps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER_Z | T_B32, 0x58); }
-void vaddsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER_Z | T_N8, 0x58); }
-void vaddss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER_Z | T_N4, 0x58); }
+void vaddsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0x58); }
+void vaddss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0x58); }
 void vaddsubpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_66 | T_0F | T_YMM, 0xD0); }
 void vaddsubps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_F2 | T_0F | T_YMM, 0xD0); }
-void vaesdec(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_66 | T_0F38 | T_W0, 0xDE); }
-void vaesdeclast(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_66 | T_0F38 | T_W0, 0xDF); }
-void vaesenc(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_66 | T_0F38 | T_W0, 0xDC); }
-void vaesenclast(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_66 | T_0F38 | T_W0, 0xDD); }
-void vaesimc(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_0F38 | T_66 | T_W0, 0xDB); }
-void vaeskeygenassist(const Xmm& xm, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A, 0xDF, imm); }
+void vaesdec(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_66 | T_0F38 | T_YMM | T_EVEX, 0xDE); }
+void vaesdeclast(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_66 | T_0F38 | T_YMM | T_EVEX, 0xDF); }
+void vaesenc(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_66 | T_0F38 | T_YMM | T_EVEX, 0xDC); }
+void vaesenclast(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_66 | T_0F38 | T_YMM | T_EVEX, 0xDD); }
+void vaesimc(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_W0, 0xDB); }
+void vaeskeygenassist(const Xmm& xm, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A, 0xDF, imm); }
 void vandnpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER_Z | T_B64, 0x55); }
 void vandnps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER_Z | T_B32, 0x55); }
 void vandpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER_Z | T_B64, 0x54); }
 void vandps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER_Z | T_B32, 0x54); }
-void vblendpd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_YMM, 0x0D, imm); }
-void vblendps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_YMM, 0x0C, imm); }
+void vbcstnebf162ps(const Xmm& x, const Address& addr) { opVex(x, 0, addr, T_F3 | T_0F38 | T_W0 | T_YMM | T_B16, 0xB1); }
+void vbcstnesh2ps(const Xmm& x, const Address& addr) { opVex(x, 0, addr, T_66 | T_0F38 | T_W0 | T_YMM | T_B16, 0xB1); }
+void vblendpd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_YMM, 0x0D, imm); }
+void vblendps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_YMM, 0x0C, imm); }
 void vblendvpd(const Xmm& x1, const Xmm& x2, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x2, op, T_0F3A | T_66 | T_YMM, 0x4B, x4.getIdx() << 4); }
 void vblendvps(const Xmm& x1, const Xmm& x2, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x2, op, T_0F3A | T_66 | T_YMM, 0x4A, x4.getIdx() << 4); }
 void vbroadcastf128(const Ymm& y, const Address& addr) { opAVX_X_XM_IMM(y, addr, T_0F38 | T_66 | T_W0 | T_YMM, 0x1A); }
 void vbroadcasti128(const Ymm& y, const Address& addr) { opAVX_X_XM_IMM(y, addr, T_0F38 | T_66 | T_W0 | T_YMM, 0x5A); }
-void vbroadcastsd(const Ymm& y, const Operand& op) { if (!op.isMEM() && !(y.isYMM() && op.isXMM()) && !(y.isZMM() && op.isXMM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(y, op, T_0F38 | T_66 | T_W0 | T_YMM | T_EVEX | T_EW1 | T_N8, 0x19); }
-void vbroadcastss(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_W0 | T_YMM | T_EVEX | T_N4, 0x18); }
+void vbroadcastsd(const Ymm& y, const Operand& op) { if (!op.isMEM() && !(y.isYMM() && op.isXMM()) && !(y.isZMM() && op.isXMM())) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_XM_IMM(y, op, T_0F38 | T_66 | T_W0 | T_YMM | T_EVEX | T_EW1 | T_N8, 0x19); }
+void vbroadcastss(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_XM_IMM(x, op, T_N4 | T_66 | T_0F38 | T_W0 | T_YMM | T_EVEX, 0x18); }
 void vcmpeq_ospd(const Xmm& x1, const Xmm& x2, const Operand& op) { vcmppd(x1, x2, op, 16); }
 void vcmpeq_osps(const Xmm& x1, const Xmm& x2, const Operand& op) { vcmpps(x1, x2, op, 16); }
 void vcmpeq_ossd(const Xmm& x1, const Xmm& x2, const Operand& op) { vcmpsd(x1, x2, op, 16); }
@@ -869,10 +973,10 @@ void vcmpordpd(const Xmm& x1, const Xmm& x2, const Operand& op) { vcmppd(x1, x2,
 void vcmpordps(const Xmm& x1, const Xmm& x2, const Operand& op) { vcmpps(x1, x2, op, 7); }
 void vcmpordsd(const Xmm& x1, const Xmm& x2, const Operand& op) { vcmpsd(x1, x2, op, 7); }
 void vcmpordss(const Xmm& x1, const Xmm& x2, const Operand& op) { vcmpss(x1, x2, op, 7); }
-void vcmppd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM, 0xC2, imm); }
-void vcmpps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_0F | T_YMM, 0xC2, imm); }
-void vcmpsd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_F2 | T_0F, 0xC2, imm); }
-void vcmpss(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_F3 | T_0F, 0xC2, imm); }
+void vcmppd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM, 0xC2, imm); }
+void vcmpps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_0F | T_YMM, 0xC2, imm); }
+void vcmpsd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_F2 | T_0F, 0xC2, imm); }
+void vcmpss(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_F3 | T_0F, 0xC2, imm); }
 void vcmptrue_uspd(const Xmm& x1, const Xmm& x2, const Operand& op) { vcmppd(x1, x2, op, 31); }
 void vcmptrue_usps(const Xmm& x1, const Xmm& x2, const Operand& op) { vcmpps(x1, x2, op, 31); }
 void vcmptrue_ussd(const Xmm& x1, const Xmm& x2, const Operand& op) { vcmpsd(x1, x2, op, 31); }
@@ -889,21 +993,26 @@ void vcmpunordpd(const Xmm& x1, const Xmm& x2, const Operand& op) { vcmppd(x1, x
 void vcmpunordps(const Xmm& x1, const Xmm& x2, const Operand& op) { vcmpps(x1, x2, op, 3); }
 void vcmpunordsd(const Xmm& x1, const Xmm& x2, const Operand& op) { vcmpsd(x1, x2, op, 3); }
 void vcmpunordss(const Xmm& x1, const Xmm& x2, const Operand& op) { vcmpss(x1, x2, op, 3); }
-void vcomisd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F | T_EW1 | T_EVEX | T_SAE_X | T_N8, 0x2F); }
-void vcomiss(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_0F | T_EW0 | T_EVEX | T_SAE_X | T_N4, 0x2F); }
+void vcomisd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N8 | T_66 | T_0F | T_EW1 | T_EVEX | T_SAE_X, 0x2F); }
+void vcomiss(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N4 | T_0F | T_EW0 | T_EVEX | T_SAE_X, 0x2F); }
 void vcvtdq2pd(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_0F | T_F3 | T_YMM | T_EVEX | T_EW0 | T_B32 | T_N8 | T_N_VL, 0xE6); }
 void vcvtdq2ps(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER_Z | T_B32, 0x5B); }
+void vcvtneebf162ps(const Xmm& x, const Address& addr) { opVex(x, 0, addr, T_F3 | T_0F38 | T_W0 | T_YMM, 0xB0); }
+void vcvtneeph2ps(const Xmm& x, const Address& addr) { opVex(x, 0, addr, T_66 | T_0F38 | T_W0 | T_YMM, 0xB0); }
+void vcvtneobf162ps(const Xmm& x, const Address& addr) { opVex(x, 0, addr, T_F2 | T_0F38 | T_W0 | T_YMM, 0xB0); }
+void vcvtneoph2ps(const Xmm& x, const Address& addr) { opVex(x, 0, addr, T_0F38 | T_W0 | T_YMM, 0xB0); }
+void vcvtneps2bf16(const Xmm& x, const Operand& op, PreferredEncoding encoding = DefaultEncoding) { opCvt2(x, op, T_F3 | T_0F38 | T_EW0 | T_YMM | T_SAE_Z | T_B32 | orEvexIf(encoding), 0x72); }
 void vcvtpd2dq(const Xmm& x, const Operand& op) { opCvt2(x, op, T_0F | T_F2 | T_YMM | T_EVEX | T_EW1 | T_B64 | T_ER_Z, 0xE6); }
 void vcvtpd2ps(const Xmm& x, const Operand& op) { opCvt2(x, op, T_0F | T_66 | T_YMM | T_EVEX | T_EW1 | T_B64 | T_ER_Z, 0x5A); }
 void vcvtph2ps(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_0F38 | T_66 | T_W0 | T_EVEX | T_EW0 | T_N8 | T_N_VL | T_SAE_Y, 0x13); }
 void vcvtps2dq(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F | T_EW0 | T_YMM | T_EVEX | T_ER_Z | T_B32, 0x5B); }
 void vcvtps2pd(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_0F | T_YMM | T_EVEX | T_EW0 | T_B32 | T_N8 | T_N_VL | T_SAE_Y, 0x5A); }
-void vcvtps2ph(const Operand& op, const Xmm& x, uint8 imm) { checkCvt1(x, op); opVex(x, 0, op, T_0F3A | T_66 | T_W0 | T_EVEX | T_EW0 | T_N8 | T_N_VL | T_SAE_Y, 0x1D, imm); }
+void vcvtps2ph(const Operand& op, const Xmm& x, uint8_t imm) { checkCvt1(x, op); opVex(x, 0, op, T_0F3A | T_66 | T_W0 | T_EVEX | T_EW0 | T_N8 | T_N_VL | T_SAE_Y | T_M_K, 0x1D, imm); }
 void vcvtsd2si(const Reg32& r, const Operand& op) { opAVX_X_X_XM(Xmm(r.getIdx()), xm0, op, T_0F | T_F2 | T_W0 | T_EVEX | T_EW0 | T_N4 | T_ER_X, 0x2D); }
-void vcvtsd2ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F2 | T_0F | T_EW1 | T_EVEX | T_ER_X | T_N8, 0x5A); }
+void vcvtsd2ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_F2 | T_0F | T_EW1 | T_EVEX | T_ER_X, 0x5A); }
 void vcvtsi2sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opCvt3(x1, x2, op, T_0F | T_F2 | T_EVEX, T_W1 | T_EW1 | T_ER_X | T_N8, T_W0 | T_EW0 | T_N4, 0x2A); }
 void vcvtsi2ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opCvt3(x1, x2, op, T_0F | T_F3 | T_EVEX | T_ER_X, T_W1 | T_EW1 | T_N8, T_W0 | T_EW0 | T_N4, 0x2A); }
-void vcvtss2sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F3 | T_0F | T_EW0 | T_EVEX | T_SAE_X | T_N4, 0x5A); }
+void vcvtss2sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_F3 | T_0F | T_EW0 | T_EVEX | T_SAE_X, 0x5A); }
 void vcvtss2si(const Reg32& r, const Operand& op) { opAVX_X_X_XM(Xmm(r.getIdx()), xm0, op, T_0F | T_F3 | T_W0 | T_EVEX | T_EW0 | T_ER_X | T_N8, 0x2D); }
 void vcvttpd2dq(const Xmm& x, const Operand& op) { opCvt2(x, op, T_66 | T_0F | T_YMM | T_EVEX |T_EW1 | T_B64 | T_ER_Z, 0xE6); }
 void vcvttps2dq(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_F3 | T_0F | T_EW0 | T_YMM | T_EVEX | T_SAE_Z | T_B32, 0x5B); }
@@ -911,25 +1020,25 @@ void vcvttsd2si(const Reg32& r, const Operand& op) { opAVX_X_X_XM(Xmm(r.getIdx()
 void vcvttss2si(const Reg32& r, const Operand& op) { opAVX_X_X_XM(Xmm(r.getIdx()), xm0, op, T_0F | T_F3 | T_W0 | T_EVEX | T_EW0 | T_SAE_X | T_N8, 0x2C); }
 void vdivpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER_Z | T_B64, 0x5E); }
 void vdivps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER_Z | T_B32, 0x5E); }
-void vdivsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER_Z | T_N8, 0x5E); }
-void vdivss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER_Z | T_N4, 0x5E); }
-void vdppd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0, 0x41, imm); }
-void vdpps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_YMM, 0x40, imm); }
-void vextractf128(const Operand& op, const Ymm& y, uint8 imm) { if (!(op.isXMEM() && y.isYMM())) throw Error(ERR_BAD_COMBINATION); opVex(y, 0, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x19, imm); }
-void vextracti128(const Operand& op, const Ymm& y, uint8 imm) { if (!(op.isXMEM() && y.isYMM())) throw Error(ERR_BAD_COMBINATION); opVex(y, 0, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x39, imm); }
-void vextractps(const Operand& op, const Xmm& x, uint8 imm) { if (!((op.isREG(32) || op.isMEM()) && x.isXMM())) throw Error(ERR_BAD_COMBINATION); opVex(x, 0, op, T_0F3A | T_66 | T_W0 | T_EVEX | T_N4, 0x17, imm); }
+void vdivsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0x5E); }
+void vdivss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0x5E); }
+void vdppd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0, 0x41, imm); }
+void vdpps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_YMM, 0x40, imm); }
+void vextractf128(const Operand& op, const Ymm& y, uint8_t imm) { if (!(op.isXMEM() && y.isYMM())) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(y, 0, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x19, imm); }
+void vextracti128(const Operand& op, const Ymm& y, uint8_t imm) { if (!(op.isXMEM() && y.isYMM())) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(y, 0, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x39, imm); }
+void vextractps(const Operand& op, const Xmm& x, uint8_t imm) { if (!((op.isREG(32) || op.isMEM()) && x.isXMM())) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(x, 0, op, T_0F3A | T_66 | T_W0 | T_EVEX | T_N4, 0x17, imm); }
 void vfmadd132pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x98); }
 void vfmadd132ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x98); }
-void vfmadd132sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0x99); }
-void vfmadd132ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0x99); }
+void vfmadd132sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X, 0x99); }
+void vfmadd132ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X, 0x99); }
 void vfmadd213pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0xA8); }
 void vfmadd213ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0xA8); }
-void vfmadd213sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0xA9); }
-void vfmadd213ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0xA9); }
+void vfmadd213sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X, 0xA9); }
+void vfmadd213ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X, 0xA9); }
 void vfmadd231pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0xB8); }
 void vfmadd231ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0xB8); }
-void vfmadd231sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0xB9); }
-void vfmadd231ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0xB9); }
+void vfmadd231sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X, 0xB9); }
+void vfmadd231ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X, 0xB9); }
 void vfmaddsub132pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x96); }
 void vfmaddsub132ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x96); }
 void vfmaddsub213pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0xA6); }
@@ -938,16 +1047,16 @@ void vfmaddsub231pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X
 void vfmaddsub231ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0xB6); }
 void vfmsub132pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x9A); }
 void vfmsub132ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x9A); }
-void vfmsub132sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0x9B); }
-void vfmsub132ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0x9B); }
+void vfmsub132sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X, 0x9B); }
+void vfmsub132ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X, 0x9B); }
 void vfmsub213pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0xAA); }
 void vfmsub213ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0xAA); }
-void vfmsub213sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0xAB); }
-void vfmsub213ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0xAB); }
+void vfmsub213sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X, 0xAB); }
+void vfmsub213ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X, 0xAB); }
 void vfmsub231pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0xBA); }
 void vfmsub231ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0xBA); }
-void vfmsub231sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0xBB); }
-void vfmsub231ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0xBB); }
+void vfmsub231sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X, 0xBB); }
+void vfmsub231ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X, 0xBB); }
 void vfmsubadd132pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x97); }
 void vfmsubadd132ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x97); }
 void vfmsubadd213pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0xA7); }
@@ -956,39 +1065,42 @@ void vfmsubadd231pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X
 void vfmsubadd231ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0xB7); }
 void vfnmadd132pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x9C); }
 void vfnmadd132ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x9C); }
-void vfnmadd132sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0x9D); }
-void vfnmadd132ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0x9D); }
+void vfnmadd132sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X, 0x9D); }
+void vfnmadd132ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X, 0x9D); }
 void vfnmadd213pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0xAC); }
 void vfnmadd213ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0xAC); }
-void vfnmadd213sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0xAD); }
-void vfnmadd213ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0xAD); }
+void vfnmadd213sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X, 0xAD); }
+void vfnmadd213ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X, 0xAD); }
 void vfnmadd231pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0xBC); }
 void vfnmadd231ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0xBC); }
-void vfnmadd231sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0xBD); }
-void vfnmadd231ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0xBD); }
+void vfnmadd231sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X, 0xBD); }
+void vfnmadd231ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X, 0xBD); }
 void vfnmsub132pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x9E); }
 void vfnmsub132ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x9E); }
-void vfnmsub132sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0x9F); }
-void vfnmsub132ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0x9F); }
+void vfnmsub132sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X, 0x9F); }
+void vfnmsub132ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X, 0x9F); }
 void vfnmsub213pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0xAE); }
 void vfnmsub213ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0xAE); }
-void vfnmsub213sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0xAF); }
-void vfnmsub213ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0xAF); }
+void vfnmsub213sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X, 0xAF); }
+void vfnmsub213ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X, 0xAF); }
 void vfnmsub231pd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0xBE); }
 void vfnmsub231ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0xBE); }
-void vfnmsub231sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0xBF); }
-void vfnmsub231ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0xBF); }
-void vgatherdpd(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_W1, 0x92, 0); }
-void vgatherdps(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_W0, 0x92, 1); }
-void vgatherqpd(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_W1, 0x93, 1); }
-void vgatherqps(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_W0, 0x93, 2); }
+void vfnmsub231sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_W1 | T_EW1 | T_EVEX | T_ER_X, 0xBF); }
+void vfnmsub231ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX | T_ER_X, 0xBF); }
+void vgatherdpd(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_YMM | T_VSIB | T_W1, 0x92, 0); }
+void vgatherdps(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_YMM | T_VSIB | T_W0, 0x92, 1); }
+void vgatherqpd(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_YMM | T_VSIB | T_W1, 0x93, 1); }
+void vgatherqps(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_YMM | T_VSIB | T_W0, 0x93, 2); }
+void vgf2p8affineinvqb(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W1 | T_EW1 | T_YMM | T_EVEX | T_SAE_Z | T_B64, 0xCF, imm); }
+void vgf2p8affineqb(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W1 | T_EW1 | T_YMM | T_EVEX | T_SAE_Z | T_B64, 0xCE, imm); }
+void vgf2p8mulb(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_SAE_Z, 0xCF); }
 void vhaddpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_66 | T_0F | T_YMM, 0x7C); }
 void vhaddps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_F2 | T_0F | T_YMM, 0x7C); }
 void vhsubpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_66 | T_0F | T_YMM, 0x7D); }
 void vhsubps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_F2 | T_0F | T_YMM, 0x7D); }
-void vinsertf128(const Ymm& y1, const Ymm& y2, const Operand& op, uint8 imm) { if (!(y1.isYMM() && y2.isYMM() && op.isXMEM())) throw Error(ERR_BAD_COMBINATION); opVex(y1, &y2, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x18, imm); }
-void vinserti128(const Ymm& y1, const Ymm& y2, const Operand& op, uint8 imm) { if (!(y1.isYMM() && y2.isYMM() && op.isXMEM())) throw Error(ERR_BAD_COMBINATION); opVex(y1, &y2, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x38, imm); }
-void vinsertps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_EW0 | T_EVEX, 0x21, imm); }
+void vinsertf128(const Ymm& y1, const Ymm& y2, const Operand& op, uint8_t imm) { if (!(y1.isYMM() && y2.isYMM() && op.isXMEM())) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(y1, &y2, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x18, imm); }
+void vinserti128(const Ymm& y1, const Ymm& y2, const Operand& op, uint8_t imm) { if (!(y1.isYMM() && y2.isYMM() && op.isXMEM())) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(y1, &y2, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x38, imm); }
+void vinsertps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F3A | T_W0 | T_EW0 | T_EVEX, 0x21, imm); }
 void vlddqu(const Xmm& x, const Address& addr) { opAVX_X_X_XM(x, cvtIdx0(x), addr, T_0F | T_F2 | T_W0 | T_YMM, 0xF0); }
 void vldmxcsr(const Address& addr) { opAVX_X_X_XM(xm2, xm0, addr, T_0F, 0xAE); }
 void vmaskmovdqu(const Xmm& x1, const Xmm& x2) { opAVX_X_X_XM(x1, xm0, x2, T_0F | T_66, 0xF7); }
@@ -998,35 +1110,35 @@ void vmaskmovps(const Address& addr, const Xmm& x1, const Xmm& x2) { opAVX_X_X_X
 void vmaskmovps(const Xmm& x1, const Xmm& x2, const Address& addr) { opAVX_X_X_XM(x1, x2, addr, T_0F38 | T_66 | T_W0 | T_YMM, 0x2C); }
 void vmaxpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER_Z | T_B64, 0x5F); }
 void vmaxps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER_Z | T_B32, 0x5F); }
-void vmaxsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER_Z | T_N8, 0x5F); }
-void vmaxss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER_Z | T_N4, 0x5F); }
+void vmaxsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0x5F); }
+void vmaxss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0x5F); }
 void vminpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER_Z | T_B64, 0x5D); }
 void vminps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER_Z | T_B32, 0x5D); }
-void vminsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER_Z | T_N8, 0x5D); }
-void vminss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER_Z | T_N4, 0x5D); }
-void vmovapd(const Address& addr, const Xmm& xmm) { opAVX_X_XM_IMM(xmm, addr, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX, 0x29); }
+void vminsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0x5D); }
+void vminss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0x5D); }
+void vmovapd(const Address& addr, const Xmm& xmm) { opAVX_X_XM_IMM(xmm, addr, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX | T_M_K, 0x29); }
 void vmovapd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX, 0x28); }
-void vmovaps(const Address& addr, const Xmm& xmm) { opAVX_X_XM_IMM(xmm, addr, T_0F | T_EW0 | T_YMM | T_EVEX, 0x29); }
+void vmovaps(const Address& addr, const Xmm& xmm) { opAVX_X_XM_IMM(xmm, addr, T_0F | T_EW0 | T_YMM | T_EVEX | T_M_K, 0x29); }
 void vmovaps(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_0F | T_EW0 | T_YMM | T_EVEX, 0x28); }
-void vmovd(const Operand& op, const Xmm& x) { if (!op.isREG(32) && !op.isMEM()) throw Error(ERR_BAD_COMBINATION); opAVX_X_X_XM(x, xm0, op, T_0F | T_66 | T_W0 | T_EVEX | T_N4, 0x7E); }
-void vmovd(const Xmm& x, const Operand& op) { if (!op.isREG(32) && !op.isMEM()) throw Error(ERR_BAD_COMBINATION); opAVX_X_X_XM(x, xm0, op, T_0F | T_66 | T_W0 | T_EVEX | T_N4, 0x6E); }
-void vmovddup(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_F2 | T_0F | T_EW1 | T_YMM | T_EVEX | T_ER_X | T_ER_Y | T_ER_Z | T_DUP, 0x12); }
+void vmovd(const Operand& op, const Xmm& x) { if (!op.isREG(32) && !op.isMEM()) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_X_XM(x, xm0, op, T_0F | T_66 | T_W0 | T_EVEX | T_N4, 0x7E); }
+void vmovd(const Xmm& x, const Operand& op) { if (!op.isREG(32) && !op.isMEM()) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_X_XM(x, xm0, op, T_0F | T_66 | T_W0 | T_EVEX | T_N4, 0x6E); }
+void vmovddup(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_DUP | T_F2 | T_0F | T_EW1 | T_YMM | T_EVEX | T_ER_X | T_ER_Y | T_ER_Z, 0x12); }
 void vmovdqa(const Address& addr, const Xmm& xmm) { opAVX_X_XM_IMM(xmm, addr, T_66 | T_0F | T_YMM, 0x7F); }
 void vmovdqa(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F | T_YMM, 0x6F); }
 void vmovdqu(const Address& addr, const Xmm& xmm) { opAVX_X_XM_IMM(xmm, addr, T_F3 | T_0F | T_YMM, 0x7F); }
 void vmovdqu(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_F3 | T_0F | T_YMM, 0x6F); }
-void vmovhlps(const Xmm& x1, const Xmm& x2, const Operand& op = Operand()) { if (!op.isNone() && !op.isXMM()) throw Error(ERR_BAD_COMBINATION); opAVX_X_X_XM(x1, x2, op, T_0F | T_EVEX | T_EW0, 0x12); }
+void vmovhlps(const Xmm& x1, const Xmm& x2, const Operand& op = Operand()) { if (!op.isNone() && !op.isXMM()) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_X_XM(x1, x2, op, T_0F | T_EVEX | T_EW0, 0x12); }
 void vmovhpd(const Address& addr, const Xmm& x) { opAVX_X_X_XM(x, xm0, addr, T_0F | T_66 | T_EVEX | T_EW1 | T_N8, 0x17); }
-void vmovhpd(const Xmm& x, const Operand& op1, const Operand& op2 = Operand()) { if (!op2.isNone() && !op2.isMEM()) throw Error(ERR_BAD_COMBINATION); opAVX_X_X_XM(x, op1, op2, T_0F | T_66 | T_EVEX | T_EW1 | T_N8, 0x16); }
+void vmovhpd(const Xmm& x, const Operand& op1, const Operand& op2 = Operand()) { if (!op2.isNone() && !op2.isMEM()) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_X_XM(x, op1, op2, T_0F | T_66 | T_EVEX | T_EW1 | T_N8, 0x16); }
 void vmovhps(const Address& addr, const Xmm& x) { opAVX_X_X_XM(x, xm0, addr, T_0F | T_EVEX | T_EW0 | T_N8, 0x17); }
-void vmovhps(const Xmm& x, const Operand& op1, const Operand& op2 = Operand()) { if (!op2.isNone() && !op2.isMEM()) throw Error(ERR_BAD_COMBINATION); opAVX_X_X_XM(x, op1, op2, T_0F | T_EVEX | T_EW0 | T_N8, 0x16); }
-void vmovlhps(const Xmm& x1, const Xmm& x2, const Operand& op = Operand()) { if (!op.isNone() && !op.isXMM()) throw Error(ERR_BAD_COMBINATION); opAVX_X_X_XM(x1, x2, op, T_0F | T_EVEX | T_EW0, 0x16); }
+void vmovhps(const Xmm& x, const Operand& op1, const Operand& op2 = Operand()) { if (!op2.isNone() && !op2.isMEM()) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_X_XM(x, op1, op2, T_0F | T_EVEX | T_EW0 | T_N8, 0x16); }
+void vmovlhps(const Xmm& x1, const Xmm& x2, const Operand& op = Operand()) { if (!op.isNone() && !op.isXMM()) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_X_XM(x1, x2, op, T_0F | T_EVEX | T_EW0, 0x16); }
 void vmovlpd(const Address& addr, const Xmm& x) { opAVX_X_X_XM(x, xm0, addr, T_0F | T_66 | T_EVEX | T_EW1 | T_N8, 0x13); }
-void vmovlpd(const Xmm& x, const Operand& op1, const Operand& op2 = Operand()) { if (!op2.isNone() && !op2.isMEM()) throw Error(ERR_BAD_COMBINATION); opAVX_X_X_XM(x, op1, op2, T_0F | T_66 | T_EVEX | T_EW1 | T_N8, 0x12); }
+void vmovlpd(const Xmm& x, const Operand& op1, const Operand& op2 = Operand()) { if (!op2.isNone() && !op2.isMEM()) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_X_XM(x, op1, op2, T_0F | T_66 | T_EVEX | T_EW1 | T_N8, 0x12); }
 void vmovlps(const Address& addr, const Xmm& x) { opAVX_X_X_XM(x, xm0, addr, T_0F | T_EVEX | T_EW0 | T_N8, 0x13); }
-void vmovlps(const Xmm& x, const Operand& op1, const Operand& op2 = Operand()) { if (!op2.isNone() && !op2.isMEM()) throw Error(ERR_BAD_COMBINATION); opAVX_X_X_XM(x, op1, op2, T_0F | T_EVEX | T_EW0 | T_N8, 0x12); }
-void vmovmskpd(const Reg& r, const Xmm& x) { if (!r.isBit(i32e)) throw Error(ERR_BAD_COMBINATION); opAVX_X_X_XM(x.isXMM() ? Xmm(r.getIdx()) : Ymm(r.getIdx()), cvtIdx0(x), x, T_0F | T_66 | T_W0 | T_YMM, 0x50); }
-void vmovmskps(const Reg& r, const Xmm& x) { if (!r.isBit(i32e)) throw Error(ERR_BAD_COMBINATION); opAVX_X_X_XM(x.isXMM() ? Xmm(r.getIdx()) : Ymm(r.getIdx()), cvtIdx0(x), x, T_0F | T_W0 | T_YMM, 0x50); }
+void vmovlps(const Xmm& x, const Operand& op1, const Operand& op2 = Operand()) { if (!op2.isNone() && !op2.isMEM()) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_X_XM(x, op1, op2, T_0F | T_EVEX | T_EW0 | T_N8, 0x12); }
+void vmovmskpd(const Reg& r, const Xmm& x) { if (!r.isBit(i32e)) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_X_XM(x.isXMM() ? Xmm(r.getIdx()) : Ymm(r.getIdx()), cvtIdx0(x), x, T_0F | T_66 | T_W0 | T_YMM, 0x50); }
+void vmovmskps(const Reg& r, const Xmm& x) { if (!r.isBit(i32e)) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_X_XM(x.isXMM() ? Xmm(r.getIdx()) : Ymm(r.getIdx()), cvtIdx0(x), x, T_0F | T_W0 | T_YMM, 0x50); }
 void vmovntdq(const Address& addr, const Xmm& x) { opVex(x, 0, addr, T_0F | T_66 | T_YMM | T_EVEX | T_EW0, 0xE7); }
 void vmovntdqa(const Xmm& x, const Address& addr) { opVex(x, 0, addr, T_0F38 | T_66 | T_YMM | T_EVEX | T_EW0, 0x2A); }
 void vmovntpd(const Address& addr, const Xmm& x) { opVex(x, 0, addr, T_0F | T_66 | T_YMM | T_EVEX | T_EW1, 0x2B); }
@@ -1034,23 +1146,23 @@ void vmovntps(const Address& addr, const Xmm& x) { opVex(x, 0, addr, T_0F | T_YM
 void vmovq(const Address& addr, const Xmm& x) { opAVX_X_X_XM(x, xm0, addr, T_0F | T_66 | T_EVEX | T_EW1 | T_N8, x.getIdx() < 16 ? 0xD6 : 0x7E); }
 void vmovq(const Xmm& x, const Address& addr) { int type, code; if (x.getIdx() < 16) { type = T_0F | T_F3; code = 0x7E; } else { type = T_0F | T_66 | T_EVEX | T_EW1 | T_N8; code = 0x6E; } opAVX_X_X_XM(x, xm0, addr, type, code); }
 void vmovq(const Xmm& x1, const Xmm& x2) { opAVX_X_X_XM(x1, xm0, x2, T_0F | T_F3 | T_EVEX | T_EW1 | T_N8, 0x7E); }
-void vmovsd(const Address& addr, const Xmm& x) { opAVX_X_X_XM(x, xm0, addr, T_F2 | T_0F | T_EW1 | T_EVEX | T_N8 | T_M_K, 0x11); }
-void vmovsd(const Xmm& x, const Address& addr) { opAVX_X_X_XM(x, xm0, addr, T_F2 | T_0F | T_EW1 | T_EVEX | T_N8, 0x10); }
-void vmovsd(const Xmm& x1, const Xmm& x2, const Operand& op = Operand()) { if (!op.isNone() && !op.isXMM()) throw Error(ERR_BAD_COMBINATION); opAVX_X_X_XM(x1, x2, op, T_F2 | T_0F | T_EW1 | T_EVEX | T_N8, 0x10); }
+void vmovsd(const Address& addr, const Xmm& x) { opAVX_X_X_XM(x, xm0, addr, T_N8 | T_F2 | T_0F | T_EW1 | T_EVEX | T_M_K, 0x11); }
+void vmovsd(const Xmm& x, const Address& addr) { opAVX_X_X_XM(x, xm0, addr, T_N8 | T_F2 | T_0F | T_EW1 | T_EVEX, 0x10); }
+void vmovsd(const Xmm& x1, const Xmm& x2, const Operand& op = Operand()) { if (!op.isNone() && !op.isXMM()) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_X_XM(x1, x2, op, T_N8 | T_F2 | T_0F | T_EW1 | T_EVEX, 0x10); }
 void vmovshdup(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_F3 | T_0F | T_EW0 | T_YMM | T_EVEX, 0x16); }
 void vmovsldup(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_F3 | T_0F | T_EW0 | T_YMM | T_EVEX, 0x12); }
-void vmovss(const Address& addr, const Xmm& x) { opAVX_X_X_XM(x, xm0, addr, T_F3 | T_0F | T_EW0 | T_EVEX | T_N4 | T_M_K, 0x11); }
-void vmovss(const Xmm& x, const Address& addr) { opAVX_X_X_XM(x, xm0, addr, T_F3 | T_0F | T_EW0 | T_EVEX | T_N4, 0x10); }
-void vmovss(const Xmm& x1, const Xmm& x2, const Operand& op = Operand()) { if (!op.isNone() && !op.isXMM()) throw Error(ERR_BAD_COMBINATION); opAVX_X_X_XM(x1, x2, op, T_F3 | T_0F | T_EW0 | T_EVEX | T_N4, 0x10); }
-void vmovupd(const Address& addr, const Xmm& xmm) { opAVX_X_XM_IMM(xmm, addr, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX, 0x11); }
+void vmovss(const Address& addr, const Xmm& x) { opAVX_X_X_XM(x, xm0, addr, T_N4 | T_F3 | T_0F | T_EW0 | T_EVEX | T_M_K, 0x11); }
+void vmovss(const Xmm& x, const Address& addr) { opAVX_X_X_XM(x, xm0, addr, T_N4 | T_F3 | T_0F | T_EW0 | T_EVEX, 0x10); }
+void vmovss(const Xmm& x1, const Xmm& x2, const Operand& op = Operand()) { if (!op.isNone() && !op.isXMM()) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_X_XM(x1, x2, op, T_N4 | T_F3 | T_0F | T_EW0 | T_EVEX, 0x10); }
+void vmovupd(const Address& addr, const Xmm& xmm) { opAVX_X_XM_IMM(xmm, addr, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX | T_M_K, 0x11); }
 void vmovupd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX, 0x10); }
-void vmovups(const Address& addr, const Xmm& xmm) { opAVX_X_XM_IMM(xmm, addr, T_0F | T_EW0 | T_YMM | T_EVEX, 0x11); }
+void vmovups(const Address& addr, const Xmm& xmm) { opAVX_X_XM_IMM(xmm, addr, T_0F | T_EW0 | T_YMM | T_EVEX | T_M_K, 0x11); }
 void vmovups(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_0F | T_EW0 | T_YMM | T_EVEX, 0x10); }
-void vmpsadbw(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_YMM, 0x42, imm); }
+void vmpsadbw(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_YMM, 0x42, imm); }
 void vmulpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER_Z | T_B64, 0x59); }
 void vmulps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER_Z | T_B32, 0x59); }
-void vmulsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER_Z | T_N8, 0x59); }
-void vmulss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER_Z | T_N4, 0x59); }
+void vmulsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0x59); }
+void vmulss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0x59); }
 void vorpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER_Z | T_B64, 0x56); }
 void vorps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER_Z | T_B32, 0x56); }
 void vpabsb(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_YMM | T_EVEX, 0x1C); }
@@ -1068,51 +1180,71 @@ void vpaddsw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1,
 void vpaddusb(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM | T_EVEX, 0xDC); }
 void vpaddusw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM | T_EVEX, 0xDD); }
 void vpaddw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM | T_EVEX, 0xFD); }
-void vpalignr(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_YMM | T_EVEX, 0x0F, imm); }
+void vpalignr(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_YMM | T_EVEX, 0x0F, imm); }
 void vpand(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM, 0xDB); }
 void vpandn(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM, 0xDF); }
 void vpavgb(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM | T_EVEX, 0xE0); }
 void vpavgw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM | T_EVEX, 0xE3); }
-void vpblendd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_YMM, 0x02, imm); }
+void vpblendd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_YMM, 0x02, imm); }
 void vpblendvb(const Xmm& x1, const Xmm& x2, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x2, op, T_0F3A | T_66 | T_YMM, 0x4C, x4.getIdx() << 4); }
-void vpblendw(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_YMM, 0x0E, imm); }
-void vpbroadcastb(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_W0 | T_YMM | T_EVEX | T_N1, 0x78); }
-void vpbroadcastd(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_W0 | T_YMM | T_EVEX | T_N4, 0x58); }
-void vpbroadcastq(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_W0 | T_EW1 | T_YMM | T_EVEX | T_N8, 0x59); }
-void vpbroadcastw(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_W0 | T_YMM | T_EVEX | T_N2, 0x79); }
-void vpclmulqdq(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0, 0x44, imm); }
+void vpblendw(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_YMM, 0x0E, imm); }
+void vpbroadcastb(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_XM_IMM(x, op, T_N1 | T_66 | T_0F38 | T_W0 | T_YMM | T_EVEX, 0x78); }
+void vpbroadcastd(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_XM_IMM(x, op, T_N4 | T_66 | T_0F38 | T_W0 | T_YMM | T_EVEX, 0x58); }
+void vpbroadcastq(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_XM_IMM(x, op, T_N8 | T_66 | T_0F38 | T_W0 | T_EW1 | T_YMM | T_EVEX, 0x59); }
+void vpbroadcastw(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_XM_IMM(x, op, T_N2 | T_66 | T_0F38 | T_W0 | T_YMM | T_EVEX, 0x79); }
+void vpclmulhqhqdq(const Xmm& x1, const Xmm& x2, const Operand& op) { vpclmulqdq(x1, x2, op, 0x11); }
+void vpclmulhqlqdq(const Xmm& x1, const Xmm& x2, const Operand& op) { vpclmulqdq(x1, x2, op, 0x01); }
+void vpclmullqhqdq(const Xmm& x1, const Xmm& x2, const Operand& op) { vpclmulqdq(x1, x2, op, 0x10); }
+void vpclmullqlqdq(const Xmm& x1, const Xmm& x2, const Operand& op) { vpclmulqdq(x1, x2, op, 0x00); }
+void vpclmulqdq(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_YMM | T_EVEX, 0x44, imm); }
 void vpcmpeqb(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM, 0x74); }
 void vpcmpeqd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM, 0x76); }
 void vpcmpeqq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM, 0x29); }
 void vpcmpeqw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM, 0x75); }
-void vpcmpestri(const Xmm& xm, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A, 0x61, imm); }
-void vpcmpestrm(const Xmm& xm, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A, 0x60, imm); }
+void vpcmpestri(const Xmm& xm, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A, 0x61, imm); }
+void vpcmpestrm(const Xmm& xm, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A, 0x60, imm); }
 void vpcmpgtb(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM, 0x64); }
 void vpcmpgtd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM, 0x66); }
 void vpcmpgtq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM, 0x37); }
 void vpcmpgtw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM, 0x65); }
-void vpcmpistri(const Xmm& xm, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A, 0x63, imm); }
-void vpcmpistrm(const Xmm& xm, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A, 0x62, imm); }
-void vperm2f128(const Ymm& y1, const Ymm& y2, const Operand& op, uint8 imm) { if (!(y1.isYMM() && y2.isYMM() && op.isYMEM())) throw Error(ERR_BAD_COMBINATION); opVex(y1, &y2, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x06, imm); }
-void vperm2i128(const Ymm& y1, const Ymm& y2, const Operand& op, uint8 imm) { if (!(y1.isYMM() && y2.isYMM() && op.isYMEM())) throw Error(ERR_BAD_COMBINATION); opVex(y1, &y2, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x46, imm); }
+void vpcmpistri(const Xmm& xm, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A, 0x63, imm); }
+void vpcmpistrm(const Xmm& xm, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A, 0x62, imm); }
+void vpdpbssd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F2 | T_0F38 | T_W0 | T_YMM, 0x50); }
+void vpdpbssds(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F2 | T_0F38 | T_W0 | T_YMM, 0x51); }
+void vpdpbsud(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F3 | T_0F38 | T_W0 | T_YMM, 0x50); }
+void vpdpbsuds(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F3 | T_0F38 | T_W0 | T_YMM, 0x51); }
+void vpdpbusd(const Xmm& x1, const Xmm& x2, const Operand& op, PreferredEncoding encoding = DefaultEncoding) { opEncoding(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_SAE_Z | T_B32, 0x50, encoding); }
+void vpdpbusds(const Xmm& x1, const Xmm& x2, const Operand& op, PreferredEncoding encoding = DefaultEncoding) { opEncoding(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_SAE_Z | T_B32, 0x51, encoding); }
+void vpdpbuud(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F38 | T_W0 | T_YMM, 0x50); }
+void vpdpbuuds(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F38 | T_W0 | T_YMM, 0x51); }
+void vpdpwssd(const Xmm& x1, const Xmm& x2, const Operand& op, PreferredEncoding encoding = DefaultEncoding) { opEncoding(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_SAE_Z | T_B32, 0x52, encoding); }
+void vpdpwssds(const Xmm& x1, const Xmm& x2, const Operand& op, PreferredEncoding encoding = DefaultEncoding) { opEncoding(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_SAE_Z | T_B32, 0x53, encoding); }
+void vpdpwsud(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F3 | T_0F38 | T_W0 | T_YMM, 0xD2); }
+void vpdpwsuds(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F3 | T_0F38 | T_W0 | T_YMM, 0xD3); }
+void vpdpwusd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_YMM, 0xD2); }
+void vpdpwusds(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_YMM, 0xD3); }
+void vpdpwuud(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F38 | T_W0 | T_YMM, 0xD2); }
+void vpdpwuuds(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F38 | T_W0 | T_YMM, 0xD3); }
+void vperm2f128(const Ymm& y1, const Ymm& y2, const Operand& op, uint8_t imm) { if (!(y1.isYMM() && y2.isYMM() && op.isYMEM())) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(y1, &y2, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x06, imm); }
+void vperm2i128(const Ymm& y1, const Ymm& y2, const Operand& op, uint8_t imm) { if (!(y1.isYMM() && y2.isYMM() && op.isYMEM())) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(y1, &y2, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x46, imm); }
 void vpermd(const Ymm& y1, const Ymm& y2, const Operand& op) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x36); }
 void vpermilpd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x0D); }
-void vpermilpd(const Xmm& xm, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_EVEX | T_B64, 0x05, imm); }
+void vpermilpd(const Xmm& xm, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_EVEX | T_B64, 0x05, imm); }
 void vpermilps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x0C); }
-void vpermilps(const Xmm& xm, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_EVEX | T_B32, 0x04, imm); }
-void vpermpd(const Ymm& y, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(y, op, T_66 | T_0F3A | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x01, imm); }
+void vpermilps(const Xmm& xm, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_EVEX | T_B32, 0x04, imm); }
+void vpermpd(const Ymm& y, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(y, op, T_66 | T_0F3A | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x01, imm); }
 void vpermpd(const Ymm& y1, const Ymm& y2, const Operand& op) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x16); }
 void vpermps(const Ymm& y1, const Ymm& y2, const Operand& op) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x16); }
-void vpermq(const Ymm& y, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(y, op, T_66 | T_0F3A | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x00, imm); }
+void vpermq(const Ymm& y, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(y, op, T_66 | T_0F3A | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x00, imm); }
 void vpermq(const Ymm& y1, const Ymm& y2, const Operand& op) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F38 | T_W0 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x36); }
-void vpextrb(const Operand& op, const Xmm& x, uint8 imm) { if (!((op.isREG(8|16|i32e) || op.isMEM()) && x.isXMM())) throw Error(ERR_BAD_COMBINATION); opVex(x, 0, op, T_0F3A | T_66 | T_EVEX | T_N1, 0x14, imm); }
-void vpextrd(const Operand& op, const Xmm& x, uint8 imm) { if (!((op.isREG(32) || op.isMEM()) && x.isXMM())) throw Error(ERR_BAD_COMBINATION); opVex(x, 0, op, T_0F3A | T_66 | T_W0 | T_EVEX | T_EW0 | T_N4, 0x16, imm); }
-void vpextrq(const Operand& op, const Xmm& x, uint8 imm) { if (!((op.isREG(64) || op.isMEM()) && x.isXMM())) throw Error(ERR_BAD_COMBINATION); opVex(x, 0, op, T_0F3A | T_66 | T_W1 | T_EVEX | T_EW1 | T_N8, 0x16, imm); }
-void vpextrw(const Operand& op, const Xmm& x, uint8 imm) { if (!((op.isREG(16|i32e) || op.isMEM()) && x.isXMM())) throw Error(ERR_BAD_COMBINATION); if (op.isREG() && x.getIdx() < 16) { opAVX_X_X_XM(Xmm(op.getIdx()), xm0, x, T_0F | T_66, 0xC5, imm); } else { opVex(x, 0, op, T_0F3A | T_66 | T_EVEX | T_N2, 0x15, imm); } }
-void vpgatherdd(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_W0, 0x90, 1); }
-void vpgatherdq(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_W1, 0x90, 0); }
-void vpgatherqd(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_W0, 0x91, 2); }
-void vpgatherqq(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_W1, 0x91, 1); }
+void vpextrb(const Operand& op, const Xmm& x, uint8_t imm) { if (!((op.isREG(8|16|i32e) || op.isMEM()) && x.isXMM())) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(x, 0, op, T_0F3A | T_66 | T_EVEX | T_N1, 0x14, imm); }
+void vpextrd(const Operand& op, const Xmm& x, uint8_t imm) { if (!((op.isREG(32) || op.isMEM()) && x.isXMM())) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(x, 0, op, T_0F3A | T_66 | T_W0 | T_EVEX | T_EW0 | T_N4, 0x16, imm); }
+void vpextrq(const Operand& op, const Xmm& x, uint8_t imm) { if (!((op.isREG(64) || op.isMEM()) && x.isXMM())) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(x, 0, op, T_0F3A | T_66 | T_W1 | T_EVEX | T_EW1 | T_N8, 0x16, imm); }
+void vpextrw(const Operand& op, const Xmm& x, uint8_t imm) { if (!((op.isREG(16|i32e) || op.isMEM()) && x.isXMM())) XBYAK_THROW(ERR_BAD_COMBINATION) if (op.isREG() && x.getIdx() < 16) { opAVX_X_X_XM(Xmm(op.getIdx()), xm0, x, T_0F | T_66, 0xC5, imm); } else { opVex(x, 0, op, T_0F3A | T_66 | T_EVEX | T_N2, 0x15, imm); } }
+void vpgatherdd(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_YMM | T_VSIB | T_W0, 0x90, 1); }
+void vpgatherdq(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_YMM | T_VSIB | T_W1, 0x90, 0); }
+void vpgatherqd(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_YMM | T_VSIB | T_W0, 0x91, 2); }
+void vpgatherqq(const Xmm& x1, const Address& addr, const Xmm& x2) { opGather(x1, addr, x2, T_0F38 | T_66 | T_YMM | T_VSIB | T_W1, 0x91, 1); }
 void vphaddd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM, 0x02); }
 void vphaddsw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM, 0x03); }
 void vphaddw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM, 0x01); }
@@ -1120,10 +1252,12 @@ void vphminposuw(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66
 void vphsubd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM, 0x06); }
 void vphsubsw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM, 0x07); }
 void vphsubw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM, 0x05); }
-void vpinsrb(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { if (!(x1.isXMM() && x2.isXMM() && (op.isREG(32) || op.isMEM()))) throw Error(ERR_BAD_COMBINATION); opVex(x1, &x2, op, T_0F3A | T_66 | T_EVEX | T_N1, 0x20, imm); }
-void vpinsrd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { if (!(x1.isXMM() && x2.isXMM() && (op.isREG(32) || op.isMEM()))) throw Error(ERR_BAD_COMBINATION); opVex(x1, &x2, op, T_0F3A | T_66 | T_W0 | T_EVEX | T_EW0 | T_N4, 0x22, imm); }
-void vpinsrq(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { if (!(x1.isXMM() && x2.isXMM() && (op.isREG(64) || op.isMEM()))) throw Error(ERR_BAD_COMBINATION); opVex(x1, &x2, op, T_0F3A | T_66 | T_W1 | T_EVEX | T_EW1 | T_N8, 0x22, imm); }
-void vpinsrw(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { if (!(x1.isXMM() && x2.isXMM() && (op.isREG(32) || op.isMEM()))) throw Error(ERR_BAD_COMBINATION); opVex(x1, &x2, op, T_0F | T_66 | T_EVEX | T_N2, 0xC4, imm); }
+void vpinsrb(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { if (!(x1.isXMM() && x2.isXMM() && (op.isREG(32) || op.isMEM()))) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(x1, &x2, op, T_0F3A | T_66 | T_EVEX | T_N1, 0x20, imm); }
+void vpinsrd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { if (!(x1.isXMM() && x2.isXMM() && (op.isREG(32) || op.isMEM()))) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(x1, &x2, op, T_0F3A | T_66 | T_W0 | T_EVEX | T_EW0 | T_N4, 0x22, imm); }
+void vpinsrq(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { if (!(x1.isXMM() && x2.isXMM() && (op.isREG(64) || op.isMEM()))) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(x1, &x2, op, T_0F3A | T_66 | T_W1 | T_EVEX | T_EW1 | T_N8, 0x22, imm); }
+void vpinsrw(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { if (!(x1.isXMM() && x2.isXMM() && (op.isREG(32) || op.isMEM()))) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(x1, &x2, op, T_0F | T_66 | T_EVEX | T_N2, 0xC4, imm); }
+void vpmadd52huq(const Xmm& x1, const Xmm& x2, const Operand& op, PreferredEncoding encoding = DefaultEncoding) { opEncoding(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_B64, 0xB5, encoding); }
+void vpmadd52luq(const Xmm& x1, const Xmm& x2, const Operand& op, PreferredEncoding encoding = DefaultEncoding) { opEncoding(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_B64, 0xB4, encoding); }
 void vpmaddubsw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM | T_EVEX, 0x04); }
 void vpmaddwd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM | T_EVEX, 0xF5); }
 void vpmaskmovd(const Address& addr, const Xmm& x1, const Xmm& x2) { opAVX_X_X_XM(x2, x1, addr, T_0F38 | T_66 | T_W0 | T_YMM, 0x8E); }
@@ -1142,19 +1276,19 @@ void vpminsw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1,
 void vpminub(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM | T_EVEX, 0xDA); }
 void vpminud(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x3B); }
 void vpminuw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM | T_EVEX, 0x3A); }
-void vpmovmskb(const Reg32e& r, const Xmm& x) { if (!x.is(Operand::XMM | Operand::YMM)) throw Error(ERR_BAD_COMBINATION); opVex(x.isYMM() ? Ymm(r.getIdx()) : Xmm(r.getIdx()), 0, x, T_0F | T_66 | T_YMM, 0xD7); }
-void vpmovsxbd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_YMM | T_EVEX | T_N4 | T_N_VL, 0x21); }
-void vpmovsxbq(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_YMM | T_EVEX | T_N2 | T_N_VL, 0x22); }
-void vpmovsxbw(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_YMM | T_EVEX | T_N8 | T_N_VL, 0x20); }
-void vpmovsxdq(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_EVEX | T_N8 | T_N_VL, 0x25); }
-void vpmovsxwd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_YMM | T_EVEX | T_N8 | T_N_VL, 0x23); }
-void vpmovsxwq(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_YMM | T_EVEX | T_N4 | T_N_VL, 0x24); }
-void vpmovzxbd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_YMM | T_EVEX | T_N4 | T_N_VL, 0x31); }
-void vpmovzxbq(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_YMM | T_EVEX | T_N2 | T_N_VL, 0x32); }
-void vpmovzxbw(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_YMM | T_EVEX | T_N8 | T_N_VL, 0x30); }
-void vpmovzxdq(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_EVEX | T_N8 | T_N_VL, 0x35); }
-void vpmovzxwd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_YMM | T_EVEX | T_N8 | T_N_VL, 0x33); }
-void vpmovzxwq(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_YMM | T_EVEX | T_N4 | T_N_VL, 0x34); }
+void vpmovmskb(const Reg32e& r, const Xmm& x) { if (!x.is(Operand::XMM | Operand::YMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(x.isYMM() ? Ymm(r.getIdx()) : Xmm(r.getIdx()), 0, x, T_0F | T_66 | T_YMM, 0xD7); }
+void vpmovsxbd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N4 | T_N_VL | T_66 | T_0F38 | T_YMM | T_EVEX, 0x21); }
+void vpmovsxbq(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N2 | T_N_VL | T_66 | T_0F38 | T_YMM | T_EVEX, 0x22); }
+void vpmovsxbw(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N8 | T_N_VL | T_66 | T_0F38 | T_YMM | T_EVEX, 0x20); }
+void vpmovsxdq(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N8 | T_N_VL | T_66 | T_0F38 | T_EW0 | T_YMM | T_EVEX, 0x25); }
+void vpmovsxwd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N8 | T_N_VL | T_66 | T_0F38 | T_YMM | T_EVEX, 0x23); }
+void vpmovsxwq(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N4 | T_N_VL | T_66 | T_0F38 | T_YMM | T_EVEX, 0x24); }
+void vpmovzxbd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N4 | T_N_VL | T_66 | T_0F38 | T_YMM | T_EVEX, 0x31); }
+void vpmovzxbq(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N2 | T_N_VL | T_66 | T_0F38 | T_YMM | T_EVEX, 0x32); }
+void vpmovzxbw(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N8 | T_N_VL | T_66 | T_0F38 | T_YMM | T_EVEX, 0x30); }
+void vpmovzxdq(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N8 | T_N_VL | T_66 | T_0F38 | T_EW0 | T_YMM | T_EVEX, 0x35); }
+void vpmovzxwd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N8 | T_N_VL | T_66 | T_0F38 | T_YMM | T_EVEX, 0x33); }
+void vpmovzxwq(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N4 | T_N_VL | T_66 | T_0F38 | T_YMM | T_EVEX, 0x34); }
 void vpmuldq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x28); }
 void vpmulhrsw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM | T_EVEX, 0x0B); }
 void vpmulhuw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM | T_EVEX, 0xE4); }
@@ -1165,35 +1299,35 @@ void vpmuludq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1
 void vpor(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM, 0xEB); }
 void vpsadbw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM | T_EVEX, 0xF6); }
 void vpshufb(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM | T_EVEX, 0x00); }
-void vpshufd(const Xmm& xm, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F | T_EW0 | T_YMM | T_EVEX | T_B32, 0x70, imm); }
-void vpshufhw(const Xmm& xm, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(xm, op, T_F3 | T_0F | T_YMM | T_EVEX, 0x70, imm); }
-void vpshuflw(const Xmm& xm, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(xm, op, T_F2 | T_0F | T_YMM | T_EVEX, 0x70, imm); }
+void vpshufd(const Xmm& xm, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F | T_EW0 | T_YMM | T_EVEX | T_B32, 0x70, imm); }
+void vpshufhw(const Xmm& xm, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(xm, op, T_F3 | T_0F | T_YMM | T_EVEX, 0x70, imm); }
+void vpshuflw(const Xmm& xm, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(xm, op, T_F2 | T_0F | T_YMM | T_EVEX, 0x70, imm); }
 void vpsignb(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM, 0x08); }
 void vpsignd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM, 0x0A); }
 void vpsignw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_YMM, 0x09); }
-void vpslld(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 6), x, op, T_66 | T_0F | T_EW0 | T_YMM | T_EVEX | T_B32, 0x72, imm); }
-void vpslld(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW0 | T_YMM | T_EVEX | T_N16, 0xF2); }
-void vpslldq(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 7), x, op, T_66 | T_0F | T_YMM | T_EVEX, 0x73, imm); }
-void vpsllq(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 6), x, op, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX | T_B64, 0x73, imm); }
-void vpsllq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX | T_N16, 0xF3); }
+void vpslld(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 6), x, op, T_66 | T_0F | T_EW0 | T_YMM | T_EVEX | T_B32 | T_MEM_EVEX, 0x72, imm); }
+void vpslld(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N16 | T_66 | T_0F | T_EW0 | T_YMM | T_EVEX, 0xF2); }
+void vpslldq(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 7), x, op, T_66 | T_0F | T_YMM | T_EVEX | T_MEM_EVEX, 0x73, imm); }
+void vpsllq(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 6), x, op, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX | T_B64 | T_MEM_EVEX, 0x73, imm); }
+void vpsllq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N16 | T_66 | T_0F | T_EW1 | T_YMM | T_EVEX, 0xF3); }
 void vpsllvd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x47); }
 void vpsllvq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x47); }
-void vpsllw(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 6), x, op, T_66 | T_0F | T_YMM | T_EVEX, 0x71, imm); }
-void vpsllw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM | T_EVEX | T_N16, 0xF1); }
-void vpsrad(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 4), x, op, T_66 | T_0F | T_EW0 | T_YMM | T_EVEX | T_B32, 0x72, imm); }
-void vpsrad(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW0 | T_YMM | T_EVEX | T_N16, 0xE2); }
+void vpsllw(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 6), x, op, T_66 | T_0F | T_YMM | T_EVEX | T_MEM_EVEX, 0x71, imm); }
+void vpsllw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N16 | T_66 | T_0F | T_YMM | T_EVEX, 0xF1); }
+void vpsrad(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 4), x, op, T_66 | T_0F | T_EW0 | T_YMM | T_EVEX | T_B32 | T_MEM_EVEX, 0x72, imm); }
+void vpsrad(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N16 | T_66 | T_0F | T_EW0 | T_YMM | T_EVEX, 0xE2); }
 void vpsravd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x46); }
-void vpsraw(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 4), x, op, T_66 | T_0F | T_YMM | T_EVEX, 0x71, imm); }
-void vpsraw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM | T_EVEX | T_N16, 0xE1); }
-void vpsrld(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 2), x, op, T_66 | T_0F | T_EW0 | T_YMM | T_EVEX | T_B32, 0x72, imm); }
-void vpsrld(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW0 | T_YMM | T_EVEX | T_N16, 0xD2); }
-void vpsrldq(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 3), x, op, T_66 | T_0F | T_YMM | T_EVEX, 0x73, imm); }
-void vpsrlq(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 2), x, op, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX | T_B64, 0x73, imm); }
-void vpsrlq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX | T_N16, 0xD3); }
+void vpsraw(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 4), x, op, T_66 | T_0F | T_YMM | T_EVEX | T_MEM_EVEX, 0x71, imm); }
+void vpsraw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N16 | T_66 | T_0F | T_YMM | T_EVEX, 0xE1); }
+void vpsrld(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 2), x, op, T_66 | T_0F | T_EW0 | T_YMM | T_EVEX | T_B32 | T_MEM_EVEX, 0x72, imm); }
+void vpsrld(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N16 | T_66 | T_0F | T_EW0 | T_YMM | T_EVEX, 0xD2); }
+void vpsrldq(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 3), x, op, T_66 | T_0F | T_YMM | T_EVEX | T_MEM_EVEX, 0x73, imm); }
+void vpsrlq(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 2), x, op, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX | T_B64 | T_MEM_EVEX, 0x73, imm); }
+void vpsrlq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N16 | T_66 | T_0F | T_EW1 | T_YMM | T_EVEX, 0xD3); }
 void vpsrlvd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x45); }
 void vpsrlvq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W1 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x45); }
-void vpsrlw(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 2), x, op, T_66 | T_0F | T_YMM | T_EVEX, 0x71, imm); }
-void vpsrlw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM | T_EVEX | T_N16, 0xD1); }
+void vpsrlw(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 2), x, op, T_66 | T_0F | T_YMM | T_EVEX | T_MEM_EVEX, 0x71, imm); }
+void vpsrlw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N16 | T_66 | T_0F | T_YMM | T_EVEX, 0xD1); }
 void vpsubb(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM | T_EVEX, 0xF8); }
 void vpsubd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW0 | T_YMM | T_EVEX | T_B32, 0xFA); }
 void vpsubq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX | T_B64, 0xFB); }
@@ -1214,27 +1348,35 @@ void vpunpcklwd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(
 void vpxor(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_YMM, 0xEF); }
 void vrcpps(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_0F | T_YMM, 0x53); }
 void vrcpss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F3 | T_0F, 0x53); }
-void vroundpd(const Xmm& xm, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A | T_YMM, 0x09, imm); }
-void vroundps(const Xmm& xm, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A | T_YMM, 0x08, imm); }
-void vroundsd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0, 0x0B, imm); }
-void vroundss(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0, 0x0A, imm); }
+void vroundpd(const Xmm& xm, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A | T_YMM, 0x09, imm); }
+void vroundps(const Xmm& xm, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F3A | T_YMM, 0x08, imm); }
+void vroundsd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0, 0x0B, imm); }
+void vroundss(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0, 0x0A, imm); }
 void vrsqrtps(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_0F | T_YMM, 0x52); }
 void vrsqrtss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F3 | T_0F, 0x52); }
-void vshufpd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX | T_B64, 0xC6, imm); }
-void vshufps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_0F | T_EW0 | T_YMM | T_EVEX | T_B32, 0xC6, imm); }
+void vsha512msg1(const Ymm& y, const Xmm& x) { if (!(y.isYMM() && x.isXMM())) XBYAK_THROW(ERR_BAD_PARAMETER) opVex(y, 0, x, T_F2 | T_0F38 | T_W0 | T_YMM, 0xCC); }
+void vsha512msg2(const Ymm& y1, const Ymm& y2) { if (!(y1.isYMM() && y2.isYMM())) XBYAK_THROW(ERR_BAD_PARAMETER) opVex(y1, 0, y2, T_F2 | T_0F38 | T_W0 | T_YMM, 0xCD); }
+void vsha512rnds2(const Ymm& y1, const Ymm& y2, const Xmm& x) { if (!(y1.isYMM() && y2.isYMM() && x.isXMM())) XBYAK_THROW(ERR_BAD_PARAMETER) opVex(y1, &y2, x, T_F2 | T_0F38 | T_W0 | T_YMM, 0xCB); }
+void vshufpd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX | T_B64, 0xC6, imm); }
+void vshufps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_0F | T_EW0 | T_YMM | T_EVEX | T_B32, 0xC6, imm); }
+void vsm3msg1(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F38 | T_W0 | T_EW0 | T_EVEX, 0xDA); }
+void vsm3msg2(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_W0 | T_EW0 | T_EVEX, 0xDA); }
+void vsm3rnds2(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_EW0 | T_EVEX, 0xDE, imm); }
+void vsm4key4(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F3 | T_0F38 | T_W0 | T_EW0 | T_EVEX, 0xDA); }
+void vsm4rnds4(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F2 | T_0F38 | T_W0 | T_EW0 | T_EVEX, 0xDA); }
 void vsqrtpd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX | T_ER_Z | T_B64, 0x51); }
 void vsqrtps(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER_Z | T_B32, 0x51); }
-void vsqrtsd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F2 | T_0F | T_EW1 | T_EVEX | T_ER_X | T_N8, 0x51); }
-void vsqrtss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F3 | T_0F | T_EW0 | T_EVEX | T_ER_X | T_N4, 0x51); }
+void vsqrtsd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_F2 | T_0F | T_EW1 | T_EVEX | T_ER_X, 0x51); }
+void vsqrtss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_F3 | T_0F | T_EW0 | T_EVEX | T_ER_X, 0x51); }
 void vstmxcsr(const Address& addr) { opAVX_X_X_XM(xm3, xm0, addr, T_0F, 0xAE); }
 void vsubpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER_Z | T_B64, 0x5C); }
 void vsubps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER_Z | T_B32, 0x5C); }
-void vsubsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER_Z | T_N8, 0x5C); }
-void vsubss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER_Z | T_N4, 0x5C); }
+void vsubsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER_X | T_N8, 0x5C); }
+void vsubss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER_X | T_N4, 0x5C); }
 void vtestpd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_YMM, 0x0F); }
 void vtestps(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F38 | T_YMM, 0x0E); }
-void vucomisd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_66 | T_0F | T_EW1 | T_EVEX | T_SAE_X | T_N8, 0x2E); }
-void vucomiss(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_0F | T_EW0 | T_EVEX | T_SAE_X | T_N4, 0x2E); }
+void vucomisd(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N8 | T_66 | T_0F | T_EW1 | T_EVEX | T_SAE_X, 0x2E); }
+void vucomiss(const Xmm& xm, const Operand& op) { opAVX_X_XM_IMM(xm, op, T_N4 | T_0F | T_EW0 | T_EVEX | T_SAE_X, 0x2E); }
 void vunpckhpd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX | T_B64, 0x15); }
 void vunpckhps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_EW0 | T_YMM | T_EVEX | T_B32, 0x15); }
 void vunpcklpd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW1 | T_YMM | T_EVEX | T_B64, 0x14); }
@@ -1246,16 +1388,19 @@ void vzeroupper() { db(0xC5); db(0xF8); db(0x77); }
 void wait() { db(0x9B); }
 void wbinvd() { db(0x0F); db(0x09); }
 void wrmsr() { db(0x0F); db(0x30); }
+void xabort(uint8_t imm) { db(0xC6); db(0xF8); db(imm); }
 void xadd(const Operand& op, const Reg& reg) { opModRM(reg, op, (op.isREG() && reg.isREG() && op.getBit() == reg.getBit()), op.isMEM(), 0x0F, 0xC0 | (reg.isBit(8) ? 0 : 1)); }
+void xbegin(uint32_t rel) { db(0xC7); db(0xF8); dd(rel); }
+void xend() { db(0x0F); db(0x01); db(0xD5); }
 void xgetbv() { db(0x0F); db(0x01); db(0xD0); }
 void xlatb() { db(0xD7); }
-void xor_(const Operand& op, uint32 imm) { opRM_I(op, imm, 0x30, 6); }
+void xor_(const Operand& op, uint32_t imm) { opRM_I(op, imm, 0x30, 6); }
 void xor_(const Operand& op1, const Operand& op2) { opRM_RM(op1, op2, 0x30); }
 void xorpd(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x57, 0x66, isXMM_XMMorMEM); }
 void xorps(const Xmm& xmm, const Operand& op) { opGen(xmm, op, 0x57, 0x100, isXMM_XMMorMEM); }
 #ifdef XBYAK_ENABLE_OMITTED_OPERAND
-void vblendpd(const Xmm& x, const Operand& op, uint8 imm) { vblendpd(x, x, op, imm); }
-void vblendps(const Xmm& x, const Operand& op, uint8 imm) { vblendps(x, x, op, imm); }
+void vblendpd(const Xmm& x, const Operand& op, uint8_t imm) { vblendpd(x, x, op, imm); }
+void vblendps(const Xmm& x, const Operand& op, uint8_t imm) { vblendps(x, x, op, imm); }
 void vblendvpd(const Xmm& x1, const Operand& op, const Xmm& x4) { vblendvpd(x1, x1, op, x4); }
 void vblendvps(const Xmm& x1, const Operand& op, const Xmm& x4) { vblendvps(x1, x1, op, x4); }
 void vcmpeq_ospd(const Xmm& x, const Operand& op) { vcmpeq_ospd(x, x, op); }
@@ -1370,10 +1515,10 @@ void vcmpordpd(const Xmm& x, const Operand& op) { vcmpordpd(x, x, op); }
 void vcmpordps(const Xmm& x, const Operand& op) { vcmpordps(x, x, op); }
 void vcmpordsd(const Xmm& x, const Operand& op) { vcmpordsd(x, x, op); }
 void vcmpordss(const Xmm& x, const Operand& op) { vcmpordss(x, x, op); }
-void vcmppd(const Xmm& x, const Operand& op, uint8 imm) { vcmppd(x, x, op, imm); }
-void vcmpps(const Xmm& x, const Operand& op, uint8 imm) { vcmpps(x, x, op, imm); }
-void vcmpsd(const Xmm& x, const Operand& op, uint8 imm) { vcmpsd(x, x, op, imm); }
-void vcmpss(const Xmm& x, const Operand& op, uint8 imm) { vcmpss(x, x, op, imm); }
+void vcmppd(const Xmm& x, const Operand& op, uint8_t imm) { vcmppd(x, x, op, imm); }
+void vcmpps(const Xmm& x, const Operand& op, uint8_t imm) { vcmpps(x, x, op, imm); }
+void vcmpsd(const Xmm& x, const Operand& op, uint8_t imm) { vcmpsd(x, x, op, imm); }
+void vcmpss(const Xmm& x, const Operand& op, uint8_t imm) { vcmpss(x, x, op, imm); }
 void vcmptrue_uspd(const Xmm& x, const Operand& op) { vcmptrue_uspd(x, x, op); }
 void vcmptrue_usps(const Xmm& x, const Operand& op) { vcmptrue_usps(x, x, op); }
 void vcmptrue_ussd(const Xmm& x, const Operand& op) { vcmptrue_ussd(x, x, op); }
@@ -1394,10 +1539,10 @@ void vcvtsd2ss(const Xmm& x, const Operand& op) { vcvtsd2ss(x, x, op); }
 void vcvtsi2sd(const Xmm& x, const Operand& op) { vcvtsi2sd(x, x, op); }
 void vcvtsi2ss(const Xmm& x, const Operand& op) { vcvtsi2ss(x, x, op); }
 void vcvtss2sd(const Xmm& x, const Operand& op) { vcvtss2sd(x, x, op); }
-void vdppd(const Xmm& x, const Operand& op, uint8 imm) { vdppd(x, x, op, imm); }
-void vdpps(const Xmm& x, const Operand& op, uint8 imm) { vdpps(x, x, op, imm); }
-void vinsertps(const Xmm& x, const Operand& op, uint8 imm) { vinsertps(x, x, op, imm); }
-void vmpsadbw(const Xmm& x, const Operand& op, uint8 imm) { vmpsadbw(x, x, op, imm); }
+void vdppd(const Xmm& x, const Operand& op, uint8_t imm) { vdppd(x, x, op, imm); }
+void vdpps(const Xmm& x, const Operand& op, uint8_t imm) { vdpps(x, x, op, imm); }
+void vinsertps(const Xmm& x, const Operand& op, uint8_t imm) { vinsertps(x, x, op, imm); }
+void vmpsadbw(const Xmm& x, const Operand& op, uint8_t imm) { vmpsadbw(x, x, op, imm); }
 void vpackssdw(const Xmm& x, const Operand& op) { vpackssdw(x, x, op); }
 void vpacksswb(const Xmm& x, const Operand& op) { vpacksswb(x, x, op); }
 void vpackusdw(const Xmm& x, const Operand& op) { vpackusdw(x, x, op); }
@@ -1410,15 +1555,15 @@ void vpaddsw(const Xmm& x, const Operand& op) { vpaddsw(x, x, op); }
 void vpaddusb(const Xmm& x, const Operand& op) { vpaddusb(x, x, op); }
 void vpaddusw(const Xmm& x, const Operand& op) { vpaddusw(x, x, op); }
 void vpaddw(const Xmm& x, const Operand& op) { vpaddw(x, x, op); }
-void vpalignr(const Xmm& x, const Operand& op, uint8 imm) { vpalignr(x, x, op, imm); }
+void vpalignr(const Xmm& x, const Operand& op, uint8_t imm) { vpalignr(x, x, op, imm); }
 void vpand(const Xmm& x, const Operand& op) { vpand(x, x, op); }
 void vpandn(const Xmm& x, const Operand& op) { vpandn(x, x, op); }
 void vpavgb(const Xmm& x, const Operand& op) { vpavgb(x, x, op); }
 void vpavgw(const Xmm& x, const Operand& op) { vpavgw(x, x, op); }
-void vpblendd(const Xmm& x, const Operand& op, uint8 imm) { vpblendd(x, x, op, imm); }
+void vpblendd(const Xmm& x, const Operand& op, uint8_t imm) { vpblendd(x, x, op, imm); }
 void vpblendvb(const Xmm& x1, const Operand& op, const Xmm& x4) { vpblendvb(x1, x1, op, x4); }
-void vpblendw(const Xmm& x, const Operand& op, uint8 imm) { vpblendw(x, x, op, imm); }
-void vpclmulqdq(const Xmm& x, const Operand& op, uint8 imm) { vpclmulqdq(x, x, op, imm); }
+void vpblendw(const Xmm& x, const Operand& op, uint8_t imm) { vpblendw(x, x, op, imm); }
+void vpclmulqdq(const Xmm& x, const Operand& op, uint8_t imm) { vpclmulqdq(x, x, op, imm); }
 void vpcmpeqb(const Xmm& x, const Operand& op) { vpcmpeqb(x, x, op); }
 void vpcmpeqd(const Xmm& x, const Operand& op) { vpcmpeqd(x, x, op); }
 void vpcmpeqq(const Xmm& x, const Operand& op) { vpcmpeqq(x, x, op); }
@@ -1433,10 +1578,10 @@ void vphaddw(const Xmm& x, const Operand& op) { vphaddw(x, x, op); }
 void vphsubd(const Xmm& x, const Operand& op) { vphsubd(x, x, op); }
 void vphsubsw(const Xmm& x, const Operand& op) { vphsubsw(x, x, op); }
 void vphsubw(const Xmm& x, const Operand& op) { vphsubw(x, x, op); }
-void vpinsrb(const Xmm& x, const Operand& op, uint8 imm) { vpinsrb(x, x, op, imm); }
-void vpinsrd(const Xmm& x, const Operand& op, uint8 imm) { vpinsrd(x, x, op, imm); }
-void vpinsrq(const Xmm& x, const Operand& op, uint8 imm) { vpinsrq(x, x, op, imm); }
-void vpinsrw(const Xmm& x, const Operand& op, uint8 imm) { vpinsrw(x, x, op, imm); }
+void vpinsrb(const Xmm& x, const Operand& op, uint8_t imm) { vpinsrb(x, x, op, imm); }
+void vpinsrd(const Xmm& x, const Operand& op, uint8_t imm) { vpinsrd(x, x, op, imm); }
+void vpinsrq(const Xmm& x, const Operand& op, uint8_t imm) { vpinsrq(x, x, op, imm); }
+void vpinsrw(const Xmm& x, const Operand& op, uint8_t imm) { vpinsrw(x, x, op, imm); }
 void vpmaddubsw(const Xmm& x, const Operand& op) { vpmaddubsw(x, x, op); }
 void vpmaddwd(const Xmm& x, const Operand& op) { vpmaddwd(x, x, op); }
 void vpmaxsb(const Xmm& x, const Operand& op) { vpmaxsb(x, x, op); }
@@ -1464,23 +1609,23 @@ void vpsignb(const Xmm& x, const Operand& op) { vpsignb(x, x, op); }
 void vpsignd(const Xmm& x, const Operand& op) { vpsignd(x, x, op); }
 void vpsignw(const Xmm& x, const Operand& op) { vpsignw(x, x, op); }
 void vpslld(const Xmm& x, const Operand& op) { vpslld(x, x, op); }
-void vpslld(const Xmm& x, uint8 imm) { vpslld(x, x, imm); }
-void vpslldq(const Xmm& x, uint8 imm) { vpslldq(x, x, imm); }
+void vpslld(const Xmm& x, uint8_t imm) { vpslld(x, x, imm); }
+void vpslldq(const Xmm& x, uint8_t imm) { vpslldq(x, x, imm); }
 void vpsllq(const Xmm& x, const Operand& op) { vpsllq(x, x, op); }
-void vpsllq(const Xmm& x, uint8 imm) { vpsllq(x, x, imm); }
+void vpsllq(const Xmm& x, uint8_t imm) { vpsllq(x, x, imm); }
 void vpsllw(const Xmm& x, const Operand& op) { vpsllw(x, x, op); }
-void vpsllw(const Xmm& x, uint8 imm) { vpsllw(x, x, imm); }
+void vpsllw(const Xmm& x, uint8_t imm) { vpsllw(x, x, imm); }
 void vpsrad(const Xmm& x, const Operand& op) { vpsrad(x, x, op); }
-void vpsrad(const Xmm& x, uint8 imm) { vpsrad(x, x, imm); }
+void vpsrad(const Xmm& x, uint8_t imm) { vpsrad(x, x, imm); }
 void vpsraw(const Xmm& x, const Operand& op) { vpsraw(x, x, op); }
-void vpsraw(const Xmm& x, uint8 imm) { vpsraw(x, x, imm); }
+void vpsraw(const Xmm& x, uint8_t imm) { vpsraw(x, x, imm); }
 void vpsrld(const Xmm& x, const Operand& op) { vpsrld(x, x, op); }
-void vpsrld(const Xmm& x, uint8 imm) { vpsrld(x, x, imm); }
-void vpsrldq(const Xmm& x, uint8 imm) { vpsrldq(x, x, imm); }
+void vpsrld(const Xmm& x, uint8_t imm) { vpsrld(x, x, imm); }
+void vpsrldq(const Xmm& x, uint8_t imm) { vpsrldq(x, x, imm); }
 void vpsrlq(const Xmm& x, const Operand& op) { vpsrlq(x, x, op); }
-void vpsrlq(const Xmm& x, uint8 imm) { vpsrlq(x, x, imm); }
+void vpsrlq(const Xmm& x, uint8_t imm) { vpsrlq(x, x, imm); }
 void vpsrlw(const Xmm& x, const Operand& op) { vpsrlw(x, x, op); }
-void vpsrlw(const Xmm& x, uint8 imm) { vpsrlw(x, x, imm); }
+void vpsrlw(const Xmm& x, uint8_t imm) { vpsrlw(x, x, imm); }
 void vpsubb(const Xmm& x, const Operand& op) { vpsubb(x, x, op); }
 void vpsubd(const Xmm& x, const Operand& op) { vpsubd(x, x, op); }
 void vpsubq(const Xmm& x, const Operand& op) { vpsubq(x, x, op); }
@@ -1499,11 +1644,11 @@ void vpunpcklqdq(const Xmm& x, const Operand& op) { vpunpcklqdq(x, x, op); }
 void vpunpcklwd(const Xmm& x, const Operand& op) { vpunpcklwd(x, x, op); }
 void vpxor(const Xmm& x, const Operand& op) { vpxor(x, x, op); }
 void vrcpss(const Xmm& x, const Operand& op) { vrcpss(x, x, op); }
-void vroundsd(const Xmm& x, const Operand& op, uint8 imm) { vroundsd(x, x, op, imm); }
-void vroundss(const Xmm& x, const Operand& op, uint8 imm) { vroundss(x, x, op, imm); }
+void vroundsd(const Xmm& x, const Operand& op, uint8_t imm) { vroundsd(x, x, op, imm); }
+void vroundss(const Xmm& x, const Operand& op, uint8_t imm) { vroundss(x, x, op, imm); }
 void vrsqrtss(const Xmm& x, const Operand& op) { vrsqrtss(x, x, op); }
-void vshufpd(const Xmm& x, const Operand& op, uint8 imm) { vshufpd(x, x, op, imm); }
-void vshufps(const Xmm& x, const Operand& op, uint8 imm) { vshufps(x, x, op, imm); }
+void vshufpd(const Xmm& x, const Operand& op, uint8_t imm) { vshufpd(x, x, op, imm); }
+void vshufps(const Xmm& x, const Operand& op, uint8_t imm) { vshufps(x, x, op, imm); }
 void vsqrtsd(const Xmm& x, const Operand& op) { vsqrtsd(x, x, op); }
 void vsqrtss(const Xmm& x, const Operand& op) { vsqrtss(x, x, op); }
 void vunpckhpd(const Xmm& x, const Operand& op) { vunpckhpd(x, x, op); }
@@ -1518,18 +1663,62 @@ void jrcxz(std::string label) { opJmp(label, T_SHORT, 0xe3, 0, 0); }
 void jrcxz(const Label& label) { opJmp(label, T_SHORT, 0xe3, 0, 0); }
 void cdqe() { db(0x48); db(0x98); }
 void cqo() { db(0x48); db(0x99); }
+void cmpsq() { db(0x48); db(0xA7); }
+void popfq() { db(0x9D); }
+void pushfq() { db(0x9C); }
+void lodsq() { db(0x48); db(0xAD); }
+void movsq() { db(0x48); db(0xA5); }
+void scasq() { db(0x48); db(0xAF); }
+void stosq() { db(0x48); db(0xAB); }
+void syscall() { db(0x0F); db(0x05); }
+void sysret() { db(0x0F); db(0x07); }
+void clui() { db(0xF3); db(0x0F); db(0x01); db(0xEE); }
+void stui() { db(0xF3); db(0x0F); db(0x01); db(0xEF); }
+void testui() { db(0xF3); db(0x0F); db(0x01); db(0xED); }
+void uiret() { db(0xF3); db(0x0F); db(0x01); db(0xEC); }
 void cmpxchg16b(const Address& addr) { opModM(addr, Reg64(1), 0x0F, 0xC7); }
+void fxrstor64(const Address& addr) { opModM(addr, Reg64(1), 0x0F, 0xAE); }
 void movq(const Reg64& reg, const Mmx& mmx) { if (mmx.isXMM()) db(0x66); opModR(mmx, reg, 0x0F, 0x7E); }
 void movq(const Mmx& mmx, const Reg64& reg) { if (mmx.isXMM()) db(0x66); opModR(mmx, reg, 0x0F, 0x6E); }
-void movsxd(const Reg64& reg, const Operand& op) { if (!op.isBit(32)) throw Error(ERR_BAD_COMBINATION); opModRM(reg, op, op.isREG(), op.isMEM(), 0x63); }
-void pextrq(const Operand& op, const Xmm& xmm, uint8 imm) { if (!op.isREG(64) && !op.isMEM()) throw Error(ERR_BAD_COMBINATION); opGen(Reg64(xmm.getIdx()), op, 0x16, 0x66, 0, imm, 0x3A); }
-void pinsrq(const Xmm& xmm, const Operand& op, uint8 imm) { if (!op.isREG(64) && !op.isMEM()) throw Error(ERR_BAD_COMBINATION); opGen(Reg64(xmm.getIdx()), op, 0x22, 0x66, 0, imm, 0x3A); }
+void movsxd(const Reg64& reg, const Operand& op) { if (!op.isBit(32)) XBYAK_THROW(ERR_BAD_COMBINATION) opModRM(reg, op, op.isREG(), op.isMEM(), 0x63); }
+void pextrq(const Operand& op, const Xmm& xmm, uint8_t imm) { if (!op.isREG(64) && !op.isMEM()) XBYAK_THROW(ERR_BAD_COMBINATION) opGen(Reg64(xmm.getIdx()), op, 0x16, 0x66, 0, imm, 0x3A); }
+void pinsrq(const Xmm& xmm, const Operand& op, uint8_t imm) { if (!op.isREG(64) && !op.isMEM()) XBYAK_THROW(ERR_BAD_COMBINATION) opGen(Reg64(xmm.getIdx()), op, 0x22, 0x66, 0, imm, 0x3A); }
+void senduipi(const Reg64& r) { db(0xF3); opModR(Reg32(6), r.cvt32(), 0x0F, 0xC7); }
 void vcvtss2si(const Reg64& r, const Operand& op) { opAVX_X_X_XM(Xmm(r.getIdx()), xm0, op, T_0F | T_F3 | T_W1 | T_EVEX | T_EW1 | T_ER_X | T_N8, 0x2D); }
 void vcvttss2si(const Reg64& r, const Operand& op) { opAVX_X_X_XM(Xmm(r.getIdx()), xm0, op, T_0F | T_F3 | T_W1 | T_EVEX | T_EW1 | T_SAE_X | T_N8, 0x2C); }
 void vcvtsd2si(const Reg64& r, const Operand& op) { opAVX_X_X_XM(Xmm(r.getIdx()), xm0, op, T_0F | T_F2 | T_W1 | T_EVEX | T_EW1 | T_N4 | T_ER_X, 0x2D); }
 void vcvttsd2si(const Reg64& r, const Operand& op) { opAVX_X_X_XM(Xmm(r.getIdx()), xm0, op, T_0F | T_F2 | T_W1 | T_EVEX | T_EW1 | T_N4 | T_SAE_X, 0x2C); }
 void vmovq(const Xmm& x, const Reg64& r) { opAVX_X_X_XM(x, xm0, Xmm(r.getIdx()), T_66 | T_0F | T_W1 | T_EVEX | T_EW1, 0x6E); }
 void vmovq(const Reg64& r, const Xmm& x) { opAVX_X_X_XM(x, xm0, Xmm(r.getIdx()), T_66 | T_0F | T_W1 | T_EVEX | T_EW1, 0x7E); }
+void cmpbexadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xE6, false); }
+void cmpbxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xE2, false); }
+void cmplexadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xEE, false); }
+void cmplxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xEC, false); }
+void cmpnbexadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xE7, false); }
+void cmpnbxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xE3, false); }
+void cmpnlexadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xEF, false); }
+void cmpnlxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xED, false); }
+void cmpnoxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xE1, false); }
+void cmpnpxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xEB, false); }
+void cmpnsxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xE9, false); }
+void cmpnzxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xE5, false); }
+void cmpoxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xE0, false); }
+void cmppxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xEA, false); }
+void cmpsxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xE8, false); }
+void cmpzxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opGpr(r1, addr, r2, T_66 | T_0F38, 0xE4, false); }
+void ldtilecfg(const Address& addr) { opVex(tmm0, &tmm0, addr, T_0F38 | T_W0, 0x49); }
+void sttilecfg(const Address& addr) { opVex(tmm0, &tmm0, addr, T_66 | T_0F38 | T_W0, 0x49); }
+void tileloadd(const Tmm& tm, const Address& addr) { opAMX(tm, addr, T_F2 | T_0F38 | T_W0, 0x4b); }
+void tileloaddt1(const Tmm& tm, const Address& addr) { opAMX(tm, addr, T_66 | T_0F38 | T_W0, 0x4b); }
+void tilerelease() { db(0xc4); db(0xe2); db(0x78); db(0x49); db(0xc0); }
+void tilestored(const Address& addr, const Tmm& tm) { opVex(tm, &tmm0, addr, T_F3 | T_0F38 | T_W0, 0x4b); }
+void tilezero(const Tmm& Tmm) { opVex(Tmm, &tmm0, tmm0, T_F2 | T_0F38 | T_W0, 0x49); }
+void tdpbssd(const Tmm& x1, const Tmm& x2, const Tmm& x3) { opVex(x1, &x3, x2, T_F2 | T_0F38 | T_W0, 0x5e); }
+void tdpbsud(const Tmm& x1, const Tmm& x2, const Tmm& x3) { opVex(x1, &x3, x2, T_F3 | T_0F38 | T_W0, 0x5e); }
+void tdpbusd(const Tmm& x1, const Tmm& x2, const Tmm& x3) { opVex(x1, &x3, x2, T_66 | T_0F38 | T_W0, 0x5e); }
+void tdpbuud(const Tmm& x1, const Tmm& x2, const Tmm& x3) { opVex(x1, &x3, x2, T_0F38 | T_W0, 0x5e); }
+void tdpfp16ps(const Tmm &x1, const Tmm &x2, const Tmm &x3) { opVex(x1, &x3, x2, T_F2 | T_0F38 | T_W0, 0x5c); }
+void tdpbf16ps(const Tmm& x1, const Tmm& x2, const Tmm& x3) { opVex(x1, &x3, x2, T_F3 | T_0F38 | T_W0, 0x5c); }
 #else
 void jcxz(std::string label) { db(0x67); opJmp(label, T_SHORT, 0xe3, 0, 0); }
 void jcxz(const Label& label) { db(0x67); opJmp(label, T_SHORT, 0xe3, 0, 0); }
@@ -1541,20 +1730,23 @@ void aam() { db(0xD4); db(0x0A); }
 void aas() { db(0x3F); }
 void daa() { db(0x27); }
 void das() { db(0x2F); }
+void into() { db(0xCE); }
 void popad() { db(0x61); }
 void popfd() { db(0x9D); }
 void pusha() { db(0x60); }
 void pushad() { db(0x60); }
 void pushfd() { db(0x9C); }
 void popa() { db(0x61); }
+void lds(const Reg& reg, const Address& addr) { opLoadSeg(addr, reg, 0xC5, 0x100); }
+void les(const Reg& reg, const Address& addr) { opLoadSeg(addr, reg, 0xC4, 0x100); }
 #endif
 #ifndef XBYAK_NO_OP_NAMES
 void and(const Operand& op1, const Operand& op2) { and_(op1, op2); }
-void and(const Operand& op, uint32 imm) { and_(op, imm); }
+void and(const Operand& op, uint32_t imm) { and_(op, imm); }
 void or(const Operand& op1, const Operand& op2) { or_(op1, op2); }
-void or(const Operand& op, uint32 imm) { or_(op, imm); }
+void or(const Operand& op, uint32_t imm) { or_(op, imm); }
 void xor(const Operand& op1, const Operand& op2) { xor_(op1, op2); }
-void xor(const Operand& op, uint32 imm) { xor_(op, imm); }
+void xor(const Operand& op, uint32_t imm) { xor_(op, imm); }
 void not(const Operand& op) { not_(op); }
 #endif
 #ifndef XBYAK_DISABLE_AVX512
@@ -1571,17 +1763,17 @@ void kandnw(const Opmask& r1, const Opmask& r2, const Opmask& r3) { opVex(r1, &r
 void kandq(const Opmask& r1, const Opmask& r2, const Opmask& r3) { opVex(r1, &r2, r3, T_L1 | T_0F | T_W1, 0x41); }
 void kandw(const Opmask& r1, const Opmask& r2, const Opmask& r3) { opVex(r1, &r2, r3, T_L1 | T_0F | T_W0, 0x41); }
 void kmovb(const Address& addr, const Opmask& k) { opVex(k, 0, addr, T_L0 | T_0F | T_66 | T_W0, 0x91); }
-void kmovb(const Opmask& k, const Operand& op) { opVex(k, 0, op, T_L0 | T_0F | T_66 | T_W0, 0x90); }
+void kmovb(const Opmask& k, const Operand& op) { if (!op.isMEM() && !op.isOPMASK()) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(k, 0, op, T_L0 | T_0F | T_66 | T_W0, 0x90); }
 void kmovb(const Opmask& k, const Reg32& r) { opVex(k, 0, r, T_L0 | T_0F | T_66 | T_W0, 0x92); }
 void kmovb(const Reg32& r, const Opmask& k) { opVex(r, 0, k, T_L0 | T_0F | T_66 | T_W0, 0x93); }
 void kmovd(const Address& addr, const Opmask& k) { opVex(k, 0, addr, T_L0 | T_0F | T_66 | T_W1, 0x91); }
-void kmovd(const Opmask& k, const Operand& op) { opVex(k, 0, op, T_L0 | T_0F | T_66 | T_W1, 0x90); }
+void kmovd(const Opmask& k, const Operand& op) { if (!op.isMEM() && !op.isOPMASK()) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(k, 0, op, T_L0 | T_0F | T_66 | T_W1, 0x90); }
 void kmovd(const Opmask& k, const Reg32& r) { opVex(k, 0, r, T_L0 | T_0F | T_F2 | T_W0, 0x92); }
 void kmovd(const Reg32& r, const Opmask& k) { opVex(r, 0, k, T_L0 | T_0F | T_F2 | T_W0, 0x93); }
 void kmovq(const Address& addr, const Opmask& k) { opVex(k, 0, addr, T_L0 | T_0F | T_W1, 0x91); }
-void kmovq(const Opmask& k, const Operand& op) { opVex(k, 0, op, T_L0 | T_0F | T_W1, 0x90); }
+void kmovq(const Opmask& k, const Operand& op) { if (!op.isMEM() && !op.isOPMASK()) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(k, 0, op, T_L0 | T_0F | T_W1, 0x90); }
 void kmovw(const Address& addr, const Opmask& k) { opVex(k, 0, addr, T_L0 | T_0F | T_W0, 0x91); }
-void kmovw(const Opmask& k, const Operand& op) { opVex(k, 0, op, T_L0 | T_0F | T_W0, 0x90); }
+void kmovw(const Opmask& k, const Operand& op) { if (!op.isMEM() && !op.isOPMASK()) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(k, 0, op, T_L0 | T_0F | T_W0, 0x90); }
 void kmovw(const Opmask& k, const Reg32& r) { opVex(k, 0, r, T_L0 | T_0F | T_W0, 0x92); }
 void kmovw(const Reg32& r, const Opmask& k) { opVex(r, 0, k, T_L0 | T_0F | T_W0, 0x93); }
 void knotb(const Opmask& r1, const Opmask& r2) { opVex(r1, 0, r2, T_0F | T_66 | T_W0, 0x44); }
@@ -1596,14 +1788,14 @@ void kortestd(const Opmask& r1, const Opmask& r2) { opVex(r1, 0, r2, T_0F | T_66
 void kortestq(const Opmask& r1, const Opmask& r2) { opVex(r1, 0, r2, T_0F | T_W1, 0x98); }
 void kortestw(const Opmask& r1, const Opmask& r2) { opVex(r1, 0, r2, T_0F | T_W0, 0x98); }
 void korw(const Opmask& r1, const Opmask& r2, const Opmask& r3) { opVex(r1, &r2, r3, T_L1 | T_0F | T_W0, 0x45); }
-void kshiftlb(const Opmask& r1, const Opmask& r2, uint8 imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W0, 0x32, imm); }
-void kshiftld(const Opmask& r1, const Opmask& r2, uint8 imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W0, 0x33, imm); }
-void kshiftlq(const Opmask& r1, const Opmask& r2, uint8 imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W1, 0x33, imm); }
-void kshiftlw(const Opmask& r1, const Opmask& r2, uint8 imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W1, 0x32, imm); }
-void kshiftrb(const Opmask& r1, const Opmask& r2, uint8 imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W0, 0x30, imm); }
-void kshiftrd(const Opmask& r1, const Opmask& r2, uint8 imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W0, 0x31, imm); }
-void kshiftrq(const Opmask& r1, const Opmask& r2, uint8 imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W1, 0x31, imm); }
-void kshiftrw(const Opmask& r1, const Opmask& r2, uint8 imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W1, 0x30, imm); }
+void kshiftlb(const Opmask& r1, const Opmask& r2, uint8_t imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W0, 0x32, imm); }
+void kshiftld(const Opmask& r1, const Opmask& r2, uint8_t imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W0, 0x33, imm); }
+void kshiftlq(const Opmask& r1, const Opmask& r2, uint8_t imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W1, 0x33, imm); }
+void kshiftlw(const Opmask& r1, const Opmask& r2, uint8_t imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W1, 0x32, imm); }
+void kshiftrb(const Opmask& r1, const Opmask& r2, uint8_t imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W0, 0x30, imm); }
+void kshiftrd(const Opmask& r1, const Opmask& r2, uint8_t imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W0, 0x31, imm); }
+void kshiftrq(const Opmask& r1, const Opmask& r2, uint8_t imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W1, 0x31, imm); }
+void kshiftrw(const Opmask& r1, const Opmask& r2, uint8_t imm) { opVex(r1, 0, r2, T_66 | T_0F3A | T_W1, 0x30, imm); }
 void ktestb(const Opmask& r1, const Opmask& r2) { opVex(r1, 0, r2, T_0F | T_66 | T_W0, 0x99); }
 void ktestd(const Opmask& r1, const Opmask& r2) { opVex(r1, 0, r2, T_0F | T_66 | T_W1, 0x99); }
 void ktestq(const Opmask& r1, const Opmask& r2) { opVex(r1, 0, r2, T_0F | T_W1, 0x99); }
@@ -1623,8 +1815,10 @@ void v4fmaddps(const Zmm& z1, const Zmm& z2, const Address& addr) { opAVX_X_X_XM
 void v4fmaddss(const Xmm& x1, const Xmm& x2, const Address& addr) { opAVX_X_X_XM(x1, x2, addr, T_0F38 | T_F2 | T_EW0 | T_MUST_EVEX | T_N16, 0x9B); }
 void v4fnmaddps(const Zmm& z1, const Zmm& z2, const Address& addr) { opAVX_X_X_XM(z1, z2, addr, T_0F38 | T_F2 | T_EW0 | T_YMM | T_MUST_EVEX | T_N16, 0xAA); }
 void v4fnmaddss(const Xmm& x1, const Xmm& x2, const Address& addr) { opAVX_X_X_XM(x1, x2, addr, T_0F38 | T_F2 | T_EW0 | T_MUST_EVEX | T_N16, 0xAB); }
-void valignd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x03, imm); }
-void valignq(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX, 0x03, imm); }
+void vaddph(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_EW0 | T_YMM | T_MUST_EVEX | T_ER_Z | T_B16, 0x58); }
+void vaddsh(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_F3 | T_EW0 | T_MUST_EVEX | T_ER_X | T_N2, 0x58); }
+void valignd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x03, imm); }
+void valignq(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX, 0x03, imm); }
 void vblendmpd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x65); }
 void vblendmps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x65); }
 void vbroadcastf32x2(const Ymm& y, const Operand& op) { opAVX_X_XM_IMM(y, op, T_66 | T_0F38 | T_YMM | T_MUST_EVEX | T_EW0 | T_N8, 0x19); }
@@ -1637,97 +1831,320 @@ void vbroadcasti32x4(const Ymm& y, const Operand& op) { opAVX_X_XM_IMM(y, op, T_
 void vbroadcasti32x8(const Zmm& z, const Operand& op) { opAVX_X_XM_IMM(z, op, T_66 | T_0F38 | T_YMM | T_MUST_EVEX | T_EW0 | T_N32, 0x5B); }
 void vbroadcasti64x2(const Ymm& y, const Operand& op) { opAVX_X_XM_IMM(y, op, T_66 | T_0F38 | T_YMM | T_MUST_EVEX | T_EW1 | T_N16, 0x5A); }
 void vbroadcasti64x4(const Zmm& z, const Operand& op) { opAVX_X_XM_IMM(z, op, T_66 | T_0F38 | T_YMM | T_MUST_EVEX | T_EW1 | T_N32, 0x5B); }
-void vcmppd(const Opmask& k, const Xmm& x, const Operand& op, uint8 imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX, 0xC2, imm); }
-void vcmpps(const Opmask& k, const Xmm& x, const Operand& op, uint8 imm) { opAVX_K_X_XM(k, x, op, T_0F | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX, 0xC2, imm); }
-void vcmpsd(const Opmask& k, const Xmm& x, const Operand& op, uint8 imm) { opAVX_K_X_XM(k, x, op, T_F2 | T_0F | T_EW1 | T_SAE_Z | T_MUST_EVEX | T_N8, 0xC2, imm); }
-void vcmpss(const Opmask& k, const Xmm& x, const Operand& op, uint8 imm) { opAVX_K_X_XM(k, x, op, T_F3 | T_0F | T_EW0 | T_SAE_Z | T_MUST_EVEX | T_N4, 0xC2, imm); }
-void vcompresspd(const Operand& op, const Xmm& x) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_N8, 0x8A); }
-void vcompressps(const Operand& op, const Xmm& x) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4, 0x8A); }
+void vcmpeq_ospd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 16); }
+void vcmpeq_osps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 16); }
+void vcmpeq_ossd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 16); }
+void vcmpeq_osss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 16); }
+void vcmpeq_uqpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 8); }
+void vcmpeq_uqps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 8); }
+void vcmpeq_uqsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 8); }
+void vcmpeq_uqss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 8); }
+void vcmpeq_uspd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 24); }
+void vcmpeq_usps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 24); }
+void vcmpeq_ussd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 24); }
+void vcmpeq_usss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 24); }
+void vcmpeqpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 0); }
+void vcmpeqps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 0); }
+void vcmpeqsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 0); }
+void vcmpeqss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 0); }
+void vcmpfalse_ospd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 27); }
+void vcmpfalse_osps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 27); }
+void vcmpfalse_ossd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 27); }
+void vcmpfalse_osss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 27); }
+void vcmpfalsepd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 11); }
+void vcmpfalseps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 11); }
+void vcmpfalsesd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 11); }
+void vcmpfalsess(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 11); }
+void vcmpge_oqpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 29); }
+void vcmpge_oqps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 29); }
+void vcmpge_oqsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 29); }
+void vcmpge_oqss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 29); }
+void vcmpgepd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 13); }
+void vcmpgeps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 13); }
+void vcmpgesd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 13); }
+void vcmpgess(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 13); }
+void vcmpgt_oqpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 30); }
+void vcmpgt_oqps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 30); }
+void vcmpgt_oqsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 30); }
+void vcmpgt_oqss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 30); }
+void vcmpgtpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 14); }
+void vcmpgtps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 14); }
+void vcmpgtsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 14); }
+void vcmpgtss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 14); }
+void vcmple_oqpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 18); }
+void vcmple_oqps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 18); }
+void vcmple_oqsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 18); }
+void vcmple_oqss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 18); }
+void vcmplepd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 2); }
+void vcmpleps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 2); }
+void vcmplesd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 2); }
+void vcmpless(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 2); }
+void vcmplt_oqpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 17); }
+void vcmplt_oqps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 17); }
+void vcmplt_oqsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 17); }
+void vcmplt_oqss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 17); }
+void vcmpltpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 1); }
+void vcmpltps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 1); }
+void vcmpltsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 1); }
+void vcmpltss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 1); }
+void vcmpneq_oqpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 12); }
+void vcmpneq_oqps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 12); }
+void vcmpneq_oqsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 12); }
+void vcmpneq_oqss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 12); }
+void vcmpneq_ospd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 28); }
+void vcmpneq_osps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 28); }
+void vcmpneq_ossd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 28); }
+void vcmpneq_osss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 28); }
+void vcmpneq_uspd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 20); }
+void vcmpneq_usps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 20); }
+void vcmpneq_ussd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 20); }
+void vcmpneq_usss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 20); }
+void vcmpneqpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 4); }
+void vcmpneqps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 4); }
+void vcmpneqsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 4); }
+void vcmpneqss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 4); }
+void vcmpnge_uqpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 25); }
+void vcmpnge_uqps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 25); }
+void vcmpnge_uqsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 25); }
+void vcmpnge_uqss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 25); }
+void vcmpngepd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 9); }
+void vcmpngeps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 9); }
+void vcmpngesd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 9); }
+void vcmpngess(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 9); }
+void vcmpngt_uqpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 26); }
+void vcmpngt_uqps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 26); }
+void vcmpngt_uqsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 26); }
+void vcmpngt_uqss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 26); }
+void vcmpngtpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 10); }
+void vcmpngtps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 10); }
+void vcmpngtsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 10); }
+void vcmpngtss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 10); }
+void vcmpnle_uqpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 22); }
+void vcmpnle_uqps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 22); }
+void vcmpnle_uqsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 22); }
+void vcmpnle_uqss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 22); }
+void vcmpnlepd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 6); }
+void vcmpnleps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 6); }
+void vcmpnlesd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 6); }
+void vcmpnless(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 6); }
+void vcmpnlt_uqpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 21); }
+void vcmpnlt_uqps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 21); }
+void vcmpnlt_uqsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 21); }
+void vcmpnlt_uqss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 21); }
+void vcmpnltpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 5); }
+void vcmpnltps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 5); }
+void vcmpnltsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 5); }
+void vcmpnltss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 5); }
+void vcmpord_spd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 23); }
+void vcmpord_sps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 23); }
+void vcmpord_ssd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 23); }
+void vcmpord_sss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 23); }
+void vcmpordpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 7); }
+void vcmpordps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 7); }
+void vcmpordsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 7); }
+void vcmpordss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 7); }
+void vcmppd(const Opmask& k, const Xmm& x, const Operand& op, uint8_t imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0xC2, imm); }
+void vcmpph(const Opmask& k, const Xmm& x, const Operand& op, uint8_t imm) { opAVX_K_X_XM(k, x, op, T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B16, 0xC2, imm); }
+void vcmpps(const Opmask& k, const Xmm& x, const Operand& op, uint8_t imm) { opAVX_K_X_XM(k, x, op, T_0F | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0xC2, imm); }
+void vcmpsd(const Opmask& k, const Xmm& x, const Operand& op, uint8_t imm) { opAVX_K_X_XM(k, x, op, T_N8 | T_F2 | T_0F | T_EW1 | T_SAE_Z | T_MUST_EVEX, 0xC2, imm); }
+void vcmpsh(const Opmask& k, const Xmm& x, const Operand& op, uint8_t imm) { opAVX_K_X_XM(k, x, op, T_N2 | T_F3 | T_0F3A | T_EW0 | T_SAE_X | T_MUST_EVEX, 0xC2, imm); }
+void vcmpss(const Opmask& k, const Xmm& x, const Operand& op, uint8_t imm) { opAVX_K_X_XM(k, x, op, T_N4 | T_F3 | T_0F | T_EW0 | T_SAE_Z | T_MUST_EVEX, 0xC2, imm); }
+void vcmptrue_uspd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 31); }
+void vcmptrue_usps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 31); }
+void vcmptrue_ussd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 31); }
+void vcmptrue_usss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 31); }
+void vcmptruepd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 15); }
+void vcmptrueps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 15); }
+void vcmptruesd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 15); }
+void vcmptruess(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 15); }
+void vcmpunord_spd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 19); }
+void vcmpunord_sps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 19); }
+void vcmpunord_ssd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 19); }
+void vcmpunord_sss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 19); }
+void vcmpunordpd(const Opmask& k, const Xmm& x, const Operand& op) { vcmppd(k, x, op, 3); }
+void vcmpunordps(const Opmask& k, const Xmm& x, const Operand& op) { vcmpps(k, x, op, 3); }
+void vcmpunordsd(const Opmask& k, const Xmm& x, const Operand& op) { vcmpsd(k, x, op, 3); }
+void vcmpunordss(const Opmask& k, const Xmm& x, const Operand& op) { vcmpss(k, x, op, 3); }
+void vcomish(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_MAP5 | T_MUST_EVEX | T_EW0 | T_SAE_X | T_N2, 0x2F); }
+void vcompressb(const Operand& op, const Xmm& x) { opAVX_X_XM_IMM(x, op, T_N1 | T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX, 0x63); }
+void vcompresspd(const Operand& op, const Xmm& x) { opAVX_X_XM_IMM(x, op, T_N8 | T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX, 0x8A); }
+void vcompressps(const Operand& op, const Xmm& x) { opAVX_X_XM_IMM(x, op, T_N4 | T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX, 0x8A); }
+void vcompressw(const Operand& op, const Xmm& x) { opAVX_X_XM_IMM(x, op, T_N2 | T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX, 0x63); }
+void vcvtdq2ph(const Xmm& x, const Operand& op) { checkCvt4(x, op); opCvt(x, op, T_N16 | T_N_VL | T_MAP5 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B32, 0x5B); }
+void vcvtne2ps2bf16(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F2 | T_0F38 | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x72); }
+void vcvtpd2ph(const Xmm& x, const Operand& op) { opCvt5(x, op, T_N16 | T_N_VL | T_66 | T_MAP5 | T_EW1 | T_ER_Z | T_MUST_EVEX | T_B64, 0x5A); }
 void vcvtpd2qq(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F | T_EW1 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B64, 0x7B); }
-void vcvtpd2udq(const Xmm& x, const Operand& op) { opCvt2(x, op, T_0F | T_YMM | T_MUST_EVEX | T_EW1 | T_B64 | T_ER_Z, 0x79); }
+void vcvtpd2udq(const Xmm& x, const Operand& op) { opCvt2(x, op, T_0F | T_EW1 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B64, 0x79); }
 void vcvtpd2uqq(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F | T_EW1 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B64, 0x79); }
-void vcvtps2qq(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_66 | T_0F | T_YMM | T_MUST_EVEX | T_EW0 | T_B32 | T_N8 | T_N_VL | T_ER_Y, 0x7B); }
+void vcvtph2dq(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_N8 | T_N_VL | T_66 | T_MAP5 | T_EW0 | T_YMM | T_ER_Y | T_MUST_EVEX | T_B16, 0x5B); }
+void vcvtph2pd(const Xmm& x, const Operand& op) { if (!op.isXMM() && !op.isMEM()) XBYAK_THROW(ERR_BAD_MEM_SIZE) opVex(x, 0, op, T_N4 | T_N_VL | T_MAP5 | T_EW0 | T_YMM | T_SAE_X | T_MUST_EVEX | T_B16, 0x5A); }
+void vcvtph2psx(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_N8 | T_N_VL | T_66 | T_MAP6 | T_EW0 | T_YMM | T_SAE_Y | T_MUST_EVEX | T_B16, 0x13); }
+void vcvtph2qq(const Xmm& x, const Operand& op) { if (!op.isXMM() && !op.isMEM()) XBYAK_THROW(ERR_BAD_MEM_SIZE) opVex(x, 0, op, T_N4 | T_N_VL | T_66 | T_MAP5 | T_EW0 | T_YMM | T_ER_X | T_MUST_EVEX | T_B16, 0x7B); }
+void vcvtph2udq(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_N8 | T_N_VL | T_MAP5 | T_EW0 | T_YMM | T_ER_Y | T_MUST_EVEX | T_B16, 0x79); }
+void vcvtph2uqq(const Xmm& x, const Operand& op) { if (!op.isXMM() && !op.isMEM()) XBYAK_THROW(ERR_BAD_MEM_SIZE) opVex(x, 0, op, T_N4 | T_N_VL | T_66 | T_MAP5 | T_EW0 | T_YMM | T_ER_X | T_MUST_EVEX | T_B16, 0x79); }
+void vcvtph2uw(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_MAP5 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0x7D); }
+void vcvtph2w(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_MAP5 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0x7D); }
+void vcvtps2phx(const Xmm& x, const Operand& op) { checkCvt4(x, op); opCvt(x, op, T_N16 | T_N_VL | T_66 | T_MAP5 | T_EW0 | T_ER_Z | T_MUST_EVEX | T_B32, 0x1D); }
+void vcvtps2qq(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_N8 | T_N_VL | T_66 | T_0F | T_EW0 | T_YMM | T_ER_Y | T_MUST_EVEX | T_B32, 0x7B); }
 void vcvtps2udq(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_0F | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B32, 0x79); }
-void vcvtps2uqq(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_66 | T_0F | T_YMM | T_MUST_EVEX | T_EW0 | T_B32 | T_N8 | T_N_VL | T_ER_Y, 0x79); }
+void vcvtps2uqq(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_N8 | T_N_VL | T_66 | T_0F | T_EW0 | T_YMM | T_ER_Y | T_MUST_EVEX | T_B32, 0x79); }
 void vcvtqq2pd(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_F3 | T_0F | T_EW1 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B64, 0xE6); }
-void vcvtqq2ps(const Xmm& x, const Operand& op) { opCvt2(x, op, T_0F | T_YMM | T_MUST_EVEX | T_EW1 | T_B64 | T_ER_Z, 0x5B); }
-void vcvtsd2usi(const Reg32e& r, const Operand& op) { int type = (T_F2 | T_0F | T_MUST_EVEX | T_N8 | T_ER_X) | (r.isREG(64) ? T_EW1 : T_EW0); opAVX_X_X_XM(Xmm(r.getIdx()), xm0, op, type, 0x79); }
-void vcvtss2usi(const Reg32e& r, const Operand& op) { int type = (T_F3 | T_0F | T_MUST_EVEX | T_N4 | T_ER_X) | (r.isREG(64) ? T_EW1 : T_EW0); opAVX_X_X_XM(Xmm(r.getIdx()), xm0, op, type, 0x79); }
+void vcvtqq2ph(const Xmm& x, const Operand& op) { opCvt5(x, op, T_N16 | T_N_VL | T_MAP5 | T_EW1 | T_ER_Z | T_MUST_EVEX | T_B64, 0x5B); }
+void vcvtqq2ps(const Xmm& x, const Operand& op) { opCvt2(x, op, T_0F | T_EW1 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B64, 0x5B); }
+void vcvtsd2sh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_F2 | T_MAP5 | T_EW1 | T_ER_X | T_MUST_EVEX, 0x5A); }
+void vcvtsd2usi(const Reg32e& r, const Operand& op) { int type = (T_N8 | T_F2 | T_0F | T_ER_X | T_MUST_EVEX) | (r.isREG(64) ? T_EW1 : T_EW0); opVex(r, &xm0, op, type, 0x79); }
+void vcvtsh2sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_F3 | T_MAP5 | T_EW0 | T_SAE_X | T_MUST_EVEX, 0x5A); }
+void vcvtsh2si(const Reg32e& r, const Operand& op) { int type = (T_N2 | T_F3 | T_MAP5 | T_ER_X | T_MUST_EVEX) | (r.isREG(64) ? T_EW1 : T_EW0); opVex(r, &xm0, op, type, 0x2D); }
+void vcvtsh2ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_MAP6 | T_EW0 | T_SAE_X | T_MUST_EVEX, 0x13); }
+void vcvtsh2usi(const Reg32e& r, const Operand& op) { int type = (T_N2 | T_F3 | T_MAP5 | T_ER_X | T_MUST_EVEX) | (r.isREG(64) ? T_EW1 : T_EW0); opVex(r, &xm0, op, type, 0x79); }
+void vcvtsi2sh(const Xmm& x1, const Xmm& x2, const Operand& op) { if (!(x1.isXMM() && x2.isXMM() && op.isBit(32|64))) XBYAK_THROW(ERR_BAD_COMBINATION) int type = (T_F3 | T_MAP5 | T_ER_R | T_MUST_EVEX | T_M_K) | (op.isBit(32) ? (T_EW0 | T_N4) : (T_EW1 | T_N8)); opVex(x1, &x2, op, type, 0x2A); }
+void vcvtss2sh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_MAP5 | T_EW0 | T_ER_X | T_MUST_EVEX, 0x1D); }
+void vcvtss2usi(const Reg32e& r, const Operand& op) { int type = (T_N4 | T_F3 | T_0F | T_ER_X | T_MUST_EVEX) | (r.isREG(64) ? T_EW1 : T_EW0); opVex(r, &xm0, op, type, 0x79); }
 void vcvttpd2qq(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x7A); }
-void vcvttpd2udq(const Xmm& x, const Operand& op) { opCvt2(x, op, T_0F | T_YMM | T_MUST_EVEX | T_EW1 | T_B64 | T_SAE_Z, 0x78); }
+void vcvttpd2udq(const Xmm& x, const Operand& op) { opCvt2(x, op, T_0F | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x78); }
 void vcvttpd2uqq(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x78); }
-void vcvttps2qq(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_66 | T_0F | T_YMM | T_MUST_EVEX | T_EW0 | T_B32 | T_N8 | T_N_VL | T_SAE_Y, 0x7A); }
+void vcvttph2dq(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_N8 | T_N_VL | T_F3 | T_MAP5 | T_EW0 | T_YMM | T_SAE_Y | T_MUST_EVEX | T_B16, 0x5B); }
+void vcvttph2qq(const Xmm& x, const Operand& op) { if (!op.isXMM() && !op.isMEM()) XBYAK_THROW(ERR_BAD_MEM_SIZE) opVex(x, 0, op, T_N4 | T_N_VL | T_66 | T_MAP5 | T_EW0 | T_YMM | T_SAE_X | T_MUST_EVEX | T_B16, 0x7A); }
+void vcvttph2udq(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_N8 | T_N_VL | T_MAP5 | T_EW0 | T_YMM | T_SAE_Y | T_MUST_EVEX | T_B16, 0x78); }
+void vcvttph2uqq(const Xmm& x, const Operand& op) { if (!op.isXMM() && !op.isMEM()) XBYAK_THROW(ERR_BAD_MEM_SIZE) opVex(x, 0, op, T_N4 | T_N_VL | T_66 | T_MAP5 | T_EW0 | T_YMM | T_SAE_X | T_MUST_EVEX | T_B16, 0x78); }
+void vcvttph2uw(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_MAP5 | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B16, 0x7C); }
+void vcvttph2w(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_MAP5 | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B16, 0x7C); }
+void vcvttps2qq(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_N8 | T_N_VL | T_66 | T_0F | T_EW0 | T_YMM | T_SAE_Y | T_MUST_EVEX | T_B32, 0x7A); }
 void vcvttps2udq(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_0F | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x78); }
-void vcvttps2uqq(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_66 | T_0F | T_YMM | T_MUST_EVEX | T_EW0 | T_B32 | T_N8 | T_N_VL | T_SAE_Y, 0x78); }
-void vcvttsd2usi(const Reg32e& r, const Operand& op) { int type = (T_F2 | T_0F | T_MUST_EVEX | T_N8 | T_SAE_X) | (r.isREG(64) ? T_EW1 : T_EW0); opAVX_X_X_XM(Xmm(r.getIdx()), xm0, op, type, 0x78); }
-void vcvttss2usi(const Reg32e& r, const Operand& op) { int type = (T_F3 | T_0F | T_MUST_EVEX | T_N4 | T_SAE_X) | (r.isREG(64) ? T_EW1 : T_EW0); opAVX_X_X_XM(Xmm(r.getIdx()), xm0, op, type, 0x78); }
-void vcvtudq2pd(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_F3 | T_0F | T_YMM | T_MUST_EVEX | T_EW0 | T_B32 | T_N8 | T_N_VL, 0x7A); }
+void vcvttps2uqq(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_N8 | T_N_VL | T_66 | T_0F | T_EW0 | T_YMM | T_SAE_Y | T_MUST_EVEX | T_B32, 0x78); }
+void vcvttsd2usi(const Reg32e& r, const Operand& op) { int type = (T_N8 | T_F2 | T_0F | T_SAE_X | T_MUST_EVEX) | (r.isREG(64) ? T_EW1 : T_EW0); opVex(r, &xm0, op, type, 0x78); }
+void vcvttsh2si(const Reg32e& r, const Operand& op) { int type = (T_N2 | T_F3 | T_MAP5 | T_EW0 | T_SAE_X | T_MUST_EVEX) | (r.isREG(64) ? T_EW1 : T_EW0); opVex(r, &xm0, op, type, 0x2C); }
+void vcvttsh2usi(const Reg32e& r, const Operand& op) { int type = (T_N2 | T_F3 | T_MAP5 | T_EW0 | T_SAE_X | T_MUST_EVEX) | (r.isREG(64) ? T_EW1 : T_EW0); opVex(r, &xm0, op, type, 0x78); }
+void vcvttss2usi(const Reg32e& r, const Operand& op) { int type = (T_N4 | T_F3 | T_0F | T_SAE_X | T_MUST_EVEX) | (r.isREG(64) ? T_EW1 : T_EW0); opVex(r, &xm0, op, type, 0x78); }
+void vcvtudq2pd(const Xmm& x, const Operand& op) { checkCvt1(x, op); opVex(x, 0, op, T_N8 | T_N_VL | T_F3 | T_0F | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x7A); }
+void vcvtudq2ph(const Xmm& x, const Operand& op) { checkCvt4(x, op); opCvt(x, op, T_N16 | T_N_VL | T_F2 | T_MAP5 | T_EW0 | T_ER_Z | T_MUST_EVEX | T_B32, 0x7A); }
 void vcvtudq2ps(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_F2 | T_0F | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B32, 0x7A); }
 void vcvtuqq2pd(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_F3 | T_0F | T_EW1 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B64, 0x7A); }
-void vcvtuqq2ps(const Xmm& x, const Operand& op) { opCvt2(x, op, T_F2 | T_0F | T_YMM | T_MUST_EVEX | T_EW1 | T_B64 | T_ER_Z, 0x7A); }
+void vcvtuqq2ph(const Xmm& x, const Operand& op) { opCvt5(x, op, T_N16 | T_N_VL | T_F2 | T_MAP5 | T_EW1 | T_ER_Z | T_MUST_EVEX | T_B64, 0x7A); }
+void vcvtuqq2ps(const Xmm& x, const Operand& op) { opCvt2(x, op, T_F2 | T_0F | T_EW1 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B64, 0x7A); }
 void vcvtusi2sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opCvt3(x1, x2, op, T_F2 | T_0F | T_MUST_EVEX, T_W1 | T_EW1 | T_ER_X | T_N8, T_W0 | T_EW0 | T_N4, 0x7B); }
+void vcvtusi2sh(const Xmm& x1, const Xmm& x2, const Operand& op) { if (!(x1.isXMM() && x2.isXMM() && op.isBit(32|64))) XBYAK_THROW(ERR_BAD_COMBINATION) int type = (T_F3 | T_MAP5 | T_ER_R | T_MUST_EVEX | T_M_K) | (op.isBit(32) ? (T_EW0 | T_N4) : (T_EW1 | T_N8)); opVex(x1, &x2, op, type, 0x7B); }
 void vcvtusi2ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opCvt3(x1, x2, op, T_F3 | T_0F | T_MUST_EVEX | T_ER_X, T_W1 | T_EW1 | T_N8, T_W0 | T_EW0 | T_N4, 0x7B); }
-void vdbpsadbw(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x42, imm); }
+void vcvtuw2ph(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_F2 | T_MAP5 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0x7D); }
+void vcvtw2ph(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_F3 | T_MAP5 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0x7D); }
+void vdbpsadbw(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x42, imm); }
+void vdivph(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_EW0 | T_YMM | T_MUST_EVEX | T_ER_Z | T_B16, 0x5E); }
+void vdivsh(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_F3 | T_EW0 | T_MUST_EVEX | T_ER_X | T_N2, 0x5E); }
+void vdpbf16ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F3 | T_0F38 | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x52); }
 void vexp2pd(const Zmm& z, const Operand& op) { opAVX_X_XM_IMM(z, op, T_66 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW1 | T_B64 | T_SAE_Z, 0xC8); }
 void vexp2ps(const Zmm& z, const Operand& op) { opAVX_X_XM_IMM(z, op, T_66 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW0 | T_B32 | T_SAE_Z, 0xC8); }
-void vexpandpd(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_N8, 0x88); }
-void vexpandps(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4, 0x88); }
-void vextractf32x4(const Operand& op, const Ymm& r, uint8 imm) { if (!op.is(Operand::MEM | Operand::XMM)) throw Error(ERR_BAD_COMBINATION); opVex(r, 0, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_N16, 0x19, imm); }
-void vextractf32x8(const Operand& op, const Zmm& r, uint8 imm) { if (!op.is(Operand::MEM | Operand::YMM)) throw Error(ERR_BAD_COMBINATION); opVex(r, 0, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_N32, 0x1B, imm); }
-void vextractf64x2(const Operand& op, const Ymm& r, uint8 imm) { if (!op.is(Operand::MEM | Operand::XMM)) throw Error(ERR_BAD_COMBINATION); opVex(r, 0, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_N16, 0x19, imm); }
-void vextractf64x4(const Operand& op, const Zmm& r, uint8 imm) { if (!op.is(Operand::MEM | Operand::YMM)) throw Error(ERR_BAD_COMBINATION); opVex(r, 0, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_N32, 0x1B, imm); }
-void vextracti32x4(const Operand& op, const Ymm& r, uint8 imm) { if (!op.is(Operand::MEM | Operand::XMM)) throw Error(ERR_BAD_COMBINATION); opVex(r, 0, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_N16, 0x39, imm); }
-void vextracti32x8(const Operand& op, const Zmm& r, uint8 imm) { if (!op.is(Operand::MEM | Operand::YMM)) throw Error(ERR_BAD_COMBINATION); opVex(r, 0, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_N32, 0x3B, imm); }
-void vextracti64x2(const Operand& op, const Ymm& r, uint8 imm) { if (!op.is(Operand::MEM | Operand::XMM)) throw Error(ERR_BAD_COMBINATION); opVex(r, 0, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_N16, 0x39, imm); }
-void vextracti64x4(const Operand& op, const Zmm& r, uint8 imm) { if (!op.is(Operand::MEM | Operand::YMM)) throw Error(ERR_BAD_COMBINATION); opVex(r, 0, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_N32, 0x3B, imm); }
-void vfixupimmpd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x54, imm); }
-void vfixupimmps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x54, imm); }
-void vfixupimmsd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_SAE_Z | T_MUST_EVEX | T_N8, 0x55, imm); }
-void vfixupimmss(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_SAE_Z | T_MUST_EVEX | T_N4, 0x55, imm); }
-void vfpclasspd(const Opmask& k, const Operand& op, uint8 imm) { if (!op.isBit(128|256|512)) throw Error(ERR_BAD_MEM_SIZE); Reg x = k; x.setBit(op.getBit()); opVex(x, 0, op, T_66 | T_0F3A | T_MUST_EVEX | T_YMM | T_EW1 | T_B64, 0x66, imm); }
-void vfpclassps(const Opmask& k, const Operand& op, uint8 imm) { if (!op.isBit(128|256|512)) throw Error(ERR_BAD_MEM_SIZE); Reg x = k; x.setBit(op.getBit()); opVex(x, 0, op, T_66 | T_0F3A | T_MUST_EVEX | T_YMM | T_EW0 | T_B32, 0x66, imm); }
-void vfpclasssd(const Opmask& k, const Operand& op, uint8 imm) { if (!op.isXMEM()) throw Error(ERR_BAD_MEM_SIZE); opVex(k, 0, op, T_66 | T_0F3A | T_MUST_EVEX | T_EW1 | T_N8, 0x67, imm); }
-void vfpclassss(const Opmask& k, const Operand& op, uint8 imm) { if (!op.isXMEM()) throw Error(ERR_BAD_MEM_SIZE); opVex(k, 0, op, T_66 | T_0F3A | T_MUST_EVEX | T_EW0 | T_N4, 0x67, imm); }
-void vgatherdpd(const Xmm& x, const Address& addr) { opGather2(x, addr, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_N8, 0x92, 1); }
-void vgatherdps(const Xmm& x, const Address& addr) { opGather2(x, addr, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4, 0x92, 0); }
-void vgatherpf0dpd(const Address& addr) { opGatherFetch(addr, zm1, T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_N8 | T_M_K, 0xC6, Operand::YMM); }
-void vgatherpf0dps(const Address& addr) { opGatherFetch(addr, zm1, T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_N4 | T_M_K, 0xC6, Operand::ZMM); }
-void vgatherpf0qpd(const Address& addr) { opGatherFetch(addr, zm1, T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_N8 | T_M_K, 0xC7, Operand::ZMM); }
-void vgatherpf0qps(const Address& addr) { opGatherFetch(addr, zm1, T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_N4 | T_M_K, 0xC7, Operand::ZMM); }
-void vgatherpf1dpd(const Address& addr) { opGatherFetch(addr, zm2, T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_N8 | T_M_K, 0xC6, Operand::YMM); }
-void vgatherpf1dps(const Address& addr) { opGatherFetch(addr, zm2, T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_N4 | T_M_K, 0xC6, Operand::ZMM); }
-void vgatherpf1qpd(const Address& addr) { opGatherFetch(addr, zm2, T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_N8 | T_M_K, 0xC7, Operand::ZMM); }
-void vgatherpf1qps(const Address& addr) { opGatherFetch(addr, zm2, T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_N4 | T_M_K, 0xC7, Operand::ZMM); }
-void vgatherqpd(const Xmm& x, const Address& addr) { opGather2(x, addr, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_N8, 0x93, 0); }
-void vgatherqps(const Xmm& x, const Address& addr) { opGather2(x, addr, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4, 0x93, 2); }
+void vexpandpd(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_N8 | T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX, 0x88); }
+void vexpandps(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_N4 | T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX, 0x88); }
+void vextractf32x4(const Operand& op, const Ymm& r, uint8_t imm) { if (!op.is(Operand::MEM | Operand::XMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r, 0, op, T_N16 | T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x19, imm); }
+void vextractf32x8(const Operand& op, const Zmm& r, uint8_t imm) { if (!op.is(Operand::MEM | Operand::YMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r, 0, op, T_N32 | T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x1B, imm); }
+void vextractf64x2(const Operand& op, const Ymm& r, uint8_t imm) { if (!op.is(Operand::MEM | Operand::XMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r, 0, op, T_N16 | T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX, 0x19, imm); }
+void vextractf64x4(const Operand& op, const Zmm& r, uint8_t imm) { if (!op.is(Operand::MEM | Operand::YMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r, 0, op, T_N32 | T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX, 0x1B, imm); }
+void vextracti32x4(const Operand& op, const Ymm& r, uint8_t imm) { if (!op.is(Operand::MEM | Operand::XMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r, 0, op, T_N16 | T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x39, imm); }
+void vextracti32x8(const Operand& op, const Zmm& r, uint8_t imm) { if (!op.is(Operand::MEM | Operand::YMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r, 0, op, T_N32 | T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x3B, imm); }
+void vextracti64x2(const Operand& op, const Ymm& r, uint8_t imm) { if (!op.is(Operand::MEM | Operand::XMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r, 0, op, T_N16 | T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX, 0x39, imm); }
+void vextracti64x4(const Operand& op, const Zmm& r, uint8_t imm) { if (!op.is(Operand::MEM | Operand::YMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r, 0, op, T_N32 | T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX, 0x3B, imm); }
+void vfcmaddcph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F2 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B32, 0x56); }
+void vfcmulcph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F2 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B32, 0xD6); }
+void vfixupimmpd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x54, imm); }
+void vfixupimmps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x54, imm); }
+void vfixupimmsd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F3A | T_EW1 | T_SAE_Z | T_MUST_EVEX, 0x55, imm); }
+void vfixupimmss(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F3A | T_EW0 | T_SAE_Z | T_MUST_EVEX, 0x55, imm); }
+void vfmadd132ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0x98); }
+void vfmadd132sh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_ER_X | T_MUST_EVEX, 0x99); }
+void vfmadd213ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0xA8); }
+void vfmadd213sh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_ER_X | T_MUST_EVEX, 0xA9); }
+void vfmadd231ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0xB8); }
+void vfmadd231sh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_ER_X | T_MUST_EVEX, 0xB9); }
+void vfmaddcph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F3 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B32, 0x56); }
+void vfmaddsub132ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0x96); }
+void vfmaddsub213ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0xA6); }
+void vfmaddsub231ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0xB6); }
+void vfmsub132ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0x9A); }
+void vfmsub132sh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_ER_X | T_MUST_EVEX, 0x9B); }
+void vfmsub213ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0xAA); }
+void vfmsub213sh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_ER_X | T_MUST_EVEX, 0xAB); }
+void vfmsub231ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0xBA); }
+void vfmsub231sh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_ER_X | T_MUST_EVEX, 0xBB); }
+void vfmsubadd132ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0x97); }
+void vfmsubadd213ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0xA7); }
+void vfmsubadd231ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0xB7); }
+void vfmulcph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_F3 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B32, 0xD6); }
+void vfnmadd132ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0x9C); }
+void vfnmadd132sh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_ER_X | T_MUST_EVEX, 0x9D); }
+void vfnmadd213ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0xAC); }
+void vfnmadd213sh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_ER_X | T_MUST_EVEX, 0xAD); }
+void vfnmadd231ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0xBC); }
+void vfnmadd231sh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_ER_X | T_MUST_EVEX, 0xBD); }
+void vfnmsub132ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0x9E); }
+void vfnmsub132sh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_ER_X | T_MUST_EVEX, 0x9F); }
+void vfnmsub213ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0xAE); }
+void vfnmsub213sh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_ER_X | T_MUST_EVEX, 0xAF); }
+void vfnmsub231ph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0xBE); }
+void vfnmsub231sh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_ER_X | T_MUST_EVEX, 0xBF); }
+void vfpclasspd(const Opmask& k, const Operand& op, uint8_t imm) { if (!op.isBit(128|256|512)) XBYAK_THROW(ERR_BAD_MEM_SIZE) opVex(k.changeBit(op.getBit()), 0, op, T_66 | T_0F3A | T_MUST_EVEX | T_YMM | T_EW1 | T_B64, 0x66, imm); }
+void vfpclassph(const Opmask& k, const Operand& op, uint8_t imm) { if (!op.isBit(128|256|512)) XBYAK_THROW(ERR_BAD_MEM_SIZE) opVex(k.changeBit(op.getBit()), 0, op, T_0F3A | T_MUST_EVEX | T_YMM | T_EW0 | T_B16, 0x66, imm); }
+void vfpclassps(const Opmask& k, const Operand& op, uint8_t imm) { if (!op.isBit(128|256|512)) XBYAK_THROW(ERR_BAD_MEM_SIZE) opVex(k.changeBit(op.getBit()), 0, op, T_66 | T_0F3A | T_MUST_EVEX | T_YMM | T_EW0 | T_B32, 0x66, imm); }
+void vfpclasssd(const Opmask& k, const Operand& op, uint8_t imm) { if (!op.isXMEM()) XBYAK_THROW(ERR_BAD_MEM_SIZE) opVex(k, 0, op, T_66 | T_0F3A | T_MUST_EVEX | T_EW1 | T_N8, 0x67, imm); }
+void vfpclasssh(const Opmask& k, const Operand& op, uint8_t imm) { if (!op.isXMEM()) XBYAK_THROW(ERR_BAD_MEM_SIZE) opVex(k, 0, op, T_0F3A | T_MUST_EVEX | T_EW0 | T_N2, 0x67, imm); }
+void vfpclassss(const Opmask& k, const Operand& op, uint8_t imm) { if (!op.isXMEM()) XBYAK_THROW(ERR_BAD_MEM_SIZE) opVex(k, 0, op, T_66 | T_0F3A | T_MUST_EVEX | T_EW0 | T_N4, 0x67, imm); }
+void vgatherdpd(const Xmm& x, const Address& addr) { opGather2(x, addr, T_N8 | T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_VSIB, 0x92, 1); }
+void vgatherdps(const Xmm& x, const Address& addr) { opGather2(x, addr, T_N4 | T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_VSIB, 0x92, 0); }
+void vgatherpf0dpd(const Address& addr) { opGatherFetch(addr, zm1, T_N8 | T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC6, Operand::YMM); }
+void vgatherpf0dps(const Address& addr) { opGatherFetch(addr, zm1, T_N4 | T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC6, Operand::ZMM); }
+void vgatherpf0qpd(const Address& addr) { opGatherFetch(addr, zm1, T_N8 | T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC7, Operand::ZMM); }
+void vgatherpf0qps(const Address& addr) { opGatherFetch(addr, zm1, T_N4 | T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC7, Operand::ZMM); }
+void vgatherpf1dpd(const Address& addr) { opGatherFetch(addr, zm2, T_N8 | T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC6, Operand::YMM); }
+void vgatherpf1dps(const Address& addr) { opGatherFetch(addr, zm2, T_N4 | T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC6, Operand::ZMM); }
+void vgatherpf1qpd(const Address& addr) { opGatherFetch(addr, zm2, T_N8 | T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC7, Operand::ZMM); }
+void vgatherpf1qps(const Address& addr) { opGatherFetch(addr, zm2, T_N4 | T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC7, Operand::ZMM); }
+void vgatherqpd(const Xmm& x, const Address& addr) { opGather2(x, addr, T_N8 | T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_VSIB, 0x93, 0); }
+void vgatherqps(const Xmm& x, const Address& addr) { opGather2(x, addr, T_N4 | T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_VSIB, 0x93, 2); }
 void vgetexppd(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x42); }
+void vgetexpph(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B16, 0x42); }
 void vgetexpps(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x42); }
-void vgetexpsd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_SAE_X | T_MUST_EVEX | T_N8, 0x43); }
-void vgetexpss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_SAE_X | T_MUST_EVEX | T_N4, 0x43); }
-void vgetmantpd(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(x, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x26, imm); }
-void vgetmantps(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(x, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x26, imm); }
-void vgetmantsd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_SAE_X | T_MUST_EVEX | T_N8, 0x27, imm); }
-void vgetmantss(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_SAE_X | T_MUST_EVEX | T_N4, 0x27, imm); }
-void vinsertf32x4(const Ymm& r1, const Ymm& r2, const Operand& op, uint8 imm) {if (!(r1.getKind() == r2.getKind() && op.is(Operand::MEM | Operand::XMM))) throw Error(ERR_BAD_COMBINATION); opVex(r1, &r2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_N16, 0x18, imm); }
-void vinsertf32x8(const Zmm& r1, const Zmm& r2, const Operand& op, uint8 imm) {if (!op.is(Operand::MEM | Operand::YMM)) throw Error(ERR_BAD_COMBINATION); opVex(r1, &r2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_N32, 0x1A, imm); }
-void vinsertf64x2(const Ymm& r1, const Ymm& r2, const Operand& op, uint8 imm) {if (!(r1.getKind() == r2.getKind() && op.is(Operand::MEM | Operand::XMM))) throw Error(ERR_BAD_COMBINATION); opVex(r1, &r2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_N16, 0x18, imm); }
-void vinsertf64x4(const Zmm& r1, const Zmm& r2, const Operand& op, uint8 imm) {if (!op.is(Operand::MEM | Operand::YMM)) throw Error(ERR_BAD_COMBINATION); opVex(r1, &r2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_N32, 0x1A, imm); }
-void vinserti32x4(const Ymm& r1, const Ymm& r2, const Operand& op, uint8 imm) {if (!(r1.getKind() == r2.getKind() && op.is(Operand::MEM | Operand::XMM))) throw Error(ERR_BAD_COMBINATION); opVex(r1, &r2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_N16, 0x38, imm); }
-void vinserti32x8(const Zmm& r1, const Zmm& r2, const Operand& op, uint8 imm) {if (!op.is(Operand::MEM | Operand::YMM)) throw Error(ERR_BAD_COMBINATION); opVex(r1, &r2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_N32, 0x3A, imm); }
-void vinserti64x2(const Ymm& r1, const Ymm& r2, const Operand& op, uint8 imm) {if (!(r1.getKind() == r2.getKind() && op.is(Operand::MEM | Operand::XMM))) throw Error(ERR_BAD_COMBINATION); opVex(r1, &r2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_N16, 0x38, imm); }
-void vinserti64x4(const Zmm& r1, const Zmm& r2, const Operand& op, uint8 imm) {if (!op.is(Operand::MEM | Operand::YMM)) throw Error(ERR_BAD_COMBINATION); opVex(r1, &r2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_N32, 0x3A, imm); }
-void vmovdqa32(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_66 | T_0F | T_EW0 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX, 0x7F); }
+void vgetexpsd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_EW1 | T_SAE_X | T_MUST_EVEX, 0x43); }
+void vgetexpsh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_SAE_X | T_MUST_EVEX, 0x43); }
+void vgetexpss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_EW0 | T_SAE_X | T_MUST_EVEX, 0x43); }
+void vgetmantpd(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(x, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x26, imm); }
+void vgetmantph(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(x, op, T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B16, 0x26, imm); }
+void vgetmantps(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(x, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x26, imm); }
+void vgetmantsd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F3A | T_EW1 | T_SAE_X | T_MUST_EVEX, 0x27, imm); }
+void vgetmantsh(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_0F3A | T_EW0 | T_SAE_X | T_MUST_EVEX, 0x27, imm); }
+void vgetmantss(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F3A | T_EW0 | T_SAE_X | T_MUST_EVEX, 0x27, imm); }
+void vinsertf32x4(const Ymm& r1, const Ymm& r2, const Operand& op, uint8_t imm) {if (!(r1.getKind() == r2.getKind() && op.is(Operand::MEM | Operand::XMM))) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r1, &r2, op, T_N16 | T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x18, imm); }
+void vinsertf32x8(const Zmm& r1, const Zmm& r2, const Operand& op, uint8_t imm) {if (!op.is(Operand::MEM | Operand::YMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r1, &r2, op, T_N32 | T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x1A, imm); }
+void vinsertf64x2(const Ymm& r1, const Ymm& r2, const Operand& op, uint8_t imm) {if (!(r1.getKind() == r2.getKind() && op.is(Operand::MEM | Operand::XMM))) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r1, &r2, op, T_N16 | T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX, 0x18, imm); }
+void vinsertf64x4(const Zmm& r1, const Zmm& r2, const Operand& op, uint8_t imm) {if (!op.is(Operand::MEM | Operand::YMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r1, &r2, op, T_N32 | T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX, 0x1A, imm); }
+void vinserti32x4(const Ymm& r1, const Ymm& r2, const Operand& op, uint8_t imm) {if (!(r1.getKind() == r2.getKind() && op.is(Operand::MEM | Operand::XMM))) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r1, &r2, op, T_N16 | T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x38, imm); }
+void vinserti32x8(const Zmm& r1, const Zmm& r2, const Operand& op, uint8_t imm) {if (!op.is(Operand::MEM | Operand::YMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r1, &r2, op, T_N32 | T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x3A, imm); }
+void vinserti64x2(const Ymm& r1, const Ymm& r2, const Operand& op, uint8_t imm) {if (!(r1.getKind() == r2.getKind() && op.is(Operand::MEM | Operand::XMM))) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r1, &r2, op, T_N16 | T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX, 0x38, imm); }
+void vinserti64x4(const Zmm& r1, const Zmm& r2, const Operand& op, uint8_t imm) {if (!op.is(Operand::MEM | Operand::YMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r1, &r2, op, T_N32 | T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX, 0x3A, imm); }
+void vmaxph(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_EW0 | T_YMM | T_MUST_EVEX | T_ER_Z | T_B16, 0x5F); }
+void vmaxsh(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_F3 | T_EW0 | T_MUST_EVEX | T_ER_X | T_N2, 0x5F); }
+void vminph(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_EW0 | T_YMM | T_MUST_EVEX | T_ER_Z | T_B16, 0x5D); }
+void vminsh(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_F3 | T_EW0 | T_MUST_EVEX | T_ER_X | T_N2, 0x5D); }
+void vmovdqa32(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_66 | T_0F | T_EW0 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX | T_M_K, 0x7F); }
 void vmovdqa32(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F | T_EW0 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX, 0x6F); }
-void vmovdqa64(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_66 | T_0F | T_EW1 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX, 0x7F); }
+void vmovdqa64(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_66 | T_0F | T_EW1 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX | T_M_K, 0x7F); }
 void vmovdqa64(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F | T_EW1 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX, 0x6F); }
-void vmovdqu16(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_F2 | T_0F | T_EW1 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX, 0x7F); }
+void vmovdqu16(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_F2 | T_0F | T_EW1 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX | T_M_K, 0x7F); }
 void vmovdqu16(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_F2 | T_0F | T_EW1 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX, 0x6F); }
-void vmovdqu32(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_F3 | T_0F | T_EW0 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX, 0x7F); }
+void vmovdqu32(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_F3 | T_0F | T_EW0 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX | T_M_K, 0x7F); }
 void vmovdqu32(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_F3 | T_0F | T_EW0 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX, 0x6F); }
-void vmovdqu64(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_F3 | T_0F | T_EW1 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX, 0x7F); }
+void vmovdqu64(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_F3 | T_0F | T_EW1 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX | T_M_K, 0x7F); }
 void vmovdqu64(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_F3 | T_0F | T_EW1 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX, 0x6F); }
-void vmovdqu8(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_F2 | T_0F | T_EW0 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX, 0x7F); }
+void vmovdqu8(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_F2 | T_0F | T_EW0 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX | T_M_K, 0x7F); }
 void vmovdqu8(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_F2 | T_0F | T_EW0 | T_YMM | T_ER_X | T_ER_Y | T_ER_Z | T_MUST_EVEX, 0x6F); }
+void vmovsh(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_N2 | T_F3 | T_MAP5 | T_EW0 | T_MUST_EVEX | T_M_K, 0x11); }
+void vmovsh(const Xmm& x, const Address& addr) { opAVX_X_X_XM(x, xm0, addr, T_N2 | T_F3 | T_MAP5 | T_EW0 | T_MUST_EVEX, 0x10); }
+void vmovsh(const Xmm& x1, const Xmm& x2, const Xmm& x3) { opAVX_X_X_XM(x1, x2, x3, T_N2 | T_F3 | T_MAP5 | T_EW0 | T_MUST_EVEX, 0x10); }
+void vmovw(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_N2 | T_66 | T_MAP5 | T_MUST_EVEX, 0x7E); }
+void vmovw(const Reg32e& r, const Xmm& x) { opAVX_X_X_XM(x, xm0, r, T_N2 | T_66 | T_MAP5 | T_MUST_EVEX, 0x7E); }
+void vmovw(const Xmm& x, const Operand& op) { if (!op.isREG(32|64) && !op.isMEM()) XBYAK_THROW(ERR_BAD_COMBINATION) opAVX_X_X_XM(x, xm0, op, T_N2 | T_66 | T_MAP5 | T_MUST_EVEX, 0x6E); }
+void vmulph(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_EW0 | T_YMM | T_MUST_EVEX | T_ER_Z | T_B16, 0x59); }
+void vmulsh(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_F3 | T_EW0 | T_MUST_EVEX | T_ER_X | T_N2, 0x59); }
+void vp2intersectd(const Opmask& k, const Xmm& x, const Operand& op) { if (k.getOpmaskIdx() != 0) XBYAK_THROW(ERR_OPMASK_IS_ALREADY_SET) opAVX_K_X_XM(k, x, op, T_F2 | T_0F38 | T_YMM | T_EVEX | T_EW0 | T_B32, 0x68); }
+void vp2intersectq(const Opmask& k, const Xmm& x, const Operand& op) { if (k.getOpmaskIdx() != 0) XBYAK_THROW(ERR_OPMASK_IS_ALREADY_SET) opAVX_K_X_XM(k, x, op, T_F2 | T_0F38 | T_YMM | T_EVEX | T_EW1 | T_B64, 0x68); }
 void vp4dpwssd(const Zmm& z1, const Zmm& z2, const Address& addr) { opAVX_X_X_XM(z1, z2, addr, T_0F38 | T_F2 | T_EW0 | T_YMM | T_MUST_EVEX | T_N16, 0x52); }
 void vp4dpwssds(const Zmm& z1, const Zmm& z2, const Address& addr) { opAVX_X_X_XM(z1, z2, addr, T_0F38 | T_F2 | T_EW0 | T_YMM | T_MUST_EVEX | T_N16, 0x53); }
 void vpabsq(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_MUST_EVEX | T_EW1 | T_B64 | T_YMM, 0x1F); }
@@ -1744,8 +2161,8 @@ void vpbroadcastd(const Xmm& x, const Reg32& r) { opVex(x, 0, r, T_66 | T_0F38 |
 void vpbroadcastmb2q(const Xmm& x, const Opmask& k) { opVex(x, 0, k, T_F3 | T_0F38 | T_YMM | T_MUST_EVEX | T_EW1, 0x2A); }
 void vpbroadcastmw2d(const Xmm& x, const Opmask& k) { opVex(x, 0, k, T_F3 | T_0F38 | T_YMM | T_MUST_EVEX | T_EW0, 0x3A); }
 void vpbroadcastw(const Xmm& x, const Reg16& r) { opVex(x, 0, r, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX, 0x7B); }
-void vpcmpb(const Opmask& k, const Xmm& x, const Operand& op, uint8 imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x3F, imm); }
-void vpcmpd(const Opmask& k, const Xmm& x, const Operand& op, uint8 imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x1F, imm); }
+void vpcmpb(const Opmask& k, const Xmm& x, const Operand& op, uint8_t imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x3F, imm); }
+void vpcmpd(const Opmask& k, const Xmm& x, const Operand& op, uint8_t imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x1F, imm); }
 void vpcmpeqb(const Opmask& k, const Xmm& x, const Operand& op) { opAVX_K_X_XM(k, x, op, T_66 | T_0F | T_YMM | T_MUST_EVEX, 0x74); }
 void vpcmpeqd(const Opmask& k, const Xmm& x, const Operand& op) { opAVX_K_X_XM(k, x, op, T_66 | T_0F | T_YMM | T_MUST_EVEX | T_B32, 0x76); }
 void vpcmpeqq(const Opmask& k, const Xmm& x, const Operand& op) { opAVX_K_X_XM(k, x, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x29); }
@@ -1754,14 +2171,14 @@ void vpcmpgtb(const Opmask& k, const Xmm& x, const Operand& op) { opAVX_K_X_XM(k
 void vpcmpgtd(const Opmask& k, const Xmm& x, const Operand& op) { opAVX_K_X_XM(k, x, op, T_66 | T_0F | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x66); }
 void vpcmpgtq(const Opmask& k, const Xmm& x, const Operand& op) { opAVX_K_X_XM(k, x, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x37); }
 void vpcmpgtw(const Opmask& k, const Xmm& x, const Operand& op) { opAVX_K_X_XM(k, x, op, T_66 | T_0F | T_YMM | T_MUST_EVEX, 0x65); }
-void vpcmpq(const Opmask& k, const Xmm& x, const Operand& op, uint8 imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x1F, imm); }
-void vpcmpub(const Opmask& k, const Xmm& x, const Operand& op, uint8 imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x3E, imm); }
-void vpcmpud(const Opmask& k, const Xmm& x, const Operand& op, uint8 imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x1E, imm); }
-void vpcmpuq(const Opmask& k, const Xmm& x, const Operand& op, uint8 imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x1E, imm); }
-void vpcmpuw(const Opmask& k, const Xmm& x, const Operand& op, uint8 imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX, 0x3E, imm); }
-void vpcmpw(const Opmask& k, const Xmm& x, const Operand& op, uint8 imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX, 0x3F, imm); }
-void vpcompressd(const Operand& op, const Xmm& x) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4, 0x8B); }
-void vpcompressq(const Operand& op, const Xmm& x) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_N8, 0x8B); }
+void vpcmpq(const Opmask& k, const Xmm& x, const Operand& op, uint8_t imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x1F, imm); }
+void vpcmpub(const Opmask& k, const Xmm& x, const Operand& op, uint8_t imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX, 0x3E, imm); }
+void vpcmpud(const Opmask& k, const Xmm& x, const Operand& op, uint8_t imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x1E, imm); }
+void vpcmpuq(const Opmask& k, const Xmm& x, const Operand& op, uint8_t imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x1E, imm); }
+void vpcmpuw(const Opmask& k, const Xmm& x, const Operand& op, uint8_t imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX, 0x3E, imm); }
+void vpcmpw(const Opmask& k, const Xmm& x, const Operand& op, uint8_t imm) { opAVX_K_X_XM(k, x, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX, 0x3F, imm); }
+void vpcompressd(const Operand& op, const Xmm& x) { opAVX_X_XM_IMM(x, op, T_N4 | T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX, 0x8B); }
+void vpcompressq(const Operand& op, const Xmm& x) { opAVX_X_XM_IMM(x, op, T_N8 | T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX, 0x8B); }
 void vpconflictd(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0xC4); }
 void vpconflictq(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0xC4); }
 void vpermb(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX, 0x8D); }
@@ -1778,70 +2195,87 @@ void vpermt2ps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x
 void vpermt2q(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x7E); }
 void vpermt2w(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX, 0x7D); }
 void vpermw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX, 0x8D); }
-void vpexpandd(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4, 0x89); }
-void vpexpandq(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_N8, 0x89); }
-void vpgatherdd(const Xmm& x, const Address& addr) { opGather2(x, addr, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4, 0x90, 0); }
-void vpgatherdq(const Xmm& x, const Address& addr) { opGather2(x, addr, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_N8, 0x90, 1); }
-void vpgatherqd(const Xmm& x, const Address& addr) { opGather2(x, addr, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4, 0x91, 2); }
-void vpgatherqq(const Xmm& x, const Address& addr) { opGather2(x, addr, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_N8, 0x91, 0); }
+void vpexpandb(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_N1 | T_66 | T_0F38 | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX, 0x62); }
+void vpexpandd(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_N4 | T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX, 0x89); }
+void vpexpandq(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_N8 | T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX, 0x89); }
+void vpexpandw(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_N2 | T_66 | T_0F38 | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX, 0x62); }
+void vpgatherdd(const Xmm& x, const Address& addr) { opGather2(x, addr, T_N4 | T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_VSIB, 0x90, 0); }
+void vpgatherdq(const Xmm& x, const Address& addr) { opGather2(x, addr, T_N8 | T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_VSIB, 0x90, 1); }
+void vpgatherqd(const Xmm& x, const Address& addr) { opGather2(x, addr, T_N4 | T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_VSIB, 0x91, 2); }
+void vpgatherqq(const Xmm& x, const Address& addr) { opGather2(x, addr, T_N8 | T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_VSIB, 0x91, 0); }
 void vplzcntd(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x44); }
 void vplzcntq(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x44); }
-void vpmadd52huq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0xB5); }
-void vpmadd52luq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0xB4); }
 void vpmaxsq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x3D); }
 void vpmaxuq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x3F); }
 void vpminsq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x39); }
 void vpminuq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x3B); }
 void vpmovb2m(const Opmask& k, const Xmm& x) { opVex(k, 0, x, T_F3 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW0, 0x29); }
 void vpmovd2m(const Opmask& k, const Xmm& x) { opVex(k, 0, x, T_F3 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW0, 0x39); }
-void vpmovdb(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4 | T_N_VL, 0x31, false); }
-void vpmovdw(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N8 | T_N_VL, 0x33, true); }
+void vpmovdb(const Operand& op, const Xmm& x) { opVmov(op, x, T_N4 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x31, false); }
+void vpmovdw(const Operand& op, const Xmm& x) { opVmov(op, x, T_N8 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x33, true); }
 void vpmovm2b(const Xmm& x, const Opmask& k) { opVex(x, 0, k, T_F3 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW0, 0x28); }
 void vpmovm2d(const Xmm& x, const Opmask& k) { opVex(x, 0, k, T_F3 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW0, 0x38); }
 void vpmovm2q(const Xmm& x, const Opmask& k) { opVex(x, 0, k, T_F3 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW1, 0x38); }
 void vpmovm2w(const Xmm& x, const Opmask& k) { opVex(x, 0, k, T_F3 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW1, 0x28); }
 void vpmovq2m(const Opmask& k, const Xmm& x) { opVex(k, 0, x, T_F3 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW1, 0x39); }
-void vpmovqb(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N2 | T_N_VL, 0x32, false); }
-void vpmovqd(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N8 | T_N_VL, 0x35, true); }
-void vpmovqw(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4 | T_N_VL, 0x34, false); }
-void vpmovsdb(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4 | T_N_VL, 0x21, false); }
-void vpmovsdw(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N8 | T_N_VL, 0x23, true); }
-void vpmovsqb(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N2 | T_N_VL, 0x22, false); }
-void vpmovsqd(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N8 | T_N_VL, 0x25, true); }
-void vpmovsqw(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4 | T_N_VL, 0x24, false); }
-void vpmovswb(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N8 | T_N_VL, 0x20, true); }
-void vpmovusdb(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4 | T_N_VL, 0x11, false); }
-void vpmovusdw(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N8 | T_N_VL, 0x13, true); }
-void vpmovusqb(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N2 | T_N_VL, 0x12, false); }
-void vpmovusqd(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N8 | T_N_VL, 0x15, true); }
-void vpmovusqw(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4 | T_N_VL, 0x14, false); }
-void vpmovuswb(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N8 | T_N_VL, 0x10, true); }
+void vpmovqb(const Operand& op, const Xmm& x) { opVmov(op, x, T_N2 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x32, false); }
+void vpmovqd(const Operand& op, const Xmm& x) { opVmov(op, x, T_N8 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x35, true); }
+void vpmovqw(const Operand& op, const Xmm& x) { opVmov(op, x, T_N4 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x34, false); }
+void vpmovsdb(const Operand& op, const Xmm& x) { opVmov(op, x, T_N4 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x21, false); }
+void vpmovsdw(const Operand& op, const Xmm& x) { opVmov(op, x, T_N8 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x23, true); }
+void vpmovsqb(const Operand& op, const Xmm& x) { opVmov(op, x, T_N2 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x22, false); }
+void vpmovsqd(const Operand& op, const Xmm& x) { opVmov(op, x, T_N8 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x25, true); }
+void vpmovsqw(const Operand& op, const Xmm& x) { opVmov(op, x, T_N4 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x24, false); }
+void vpmovswb(const Operand& op, const Xmm& x) { opVmov(op, x, T_N8 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x20, true); }
+void vpmovusdb(const Operand& op, const Xmm& x) { opVmov(op, x, T_N4 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x11, false); }
+void vpmovusdw(const Operand& op, const Xmm& x) { opVmov(op, x, T_N8 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x13, true); }
+void vpmovusqb(const Operand& op, const Xmm& x) { opVmov(op, x, T_N2 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x12, false); }
+void vpmovusqd(const Operand& op, const Xmm& x) { opVmov(op, x, T_N8 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x15, true); }
+void vpmovusqw(const Operand& op, const Xmm& x) { opVmov(op, x, T_N4 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x14, false); }
+void vpmovuswb(const Operand& op, const Xmm& x) { opVmov(op, x, T_N8 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x10, true); }
 void vpmovw2m(const Opmask& k, const Xmm& x) { opVex(k, 0, x, T_F3 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW1, 0x29); }
-void vpmovwb(const Operand& op, const Xmm& x) { opVmov(op, x, T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N8 | T_N_VL, 0x30, true); }
+void vpmovwb(const Operand& op, const Xmm& x) { opVmov(op, x, T_N8 | T_N_VL | T_F3 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K, 0x30, true); }
 void vpmullq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x40); }
 void vpmultishiftqb(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x83); }
+void vpopcntb(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX, 0x54); }
+void vpopcntd(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x55); }
+void vpopcntq(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x55); }
+void vpopcntw(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX, 0x54); }
 void vpord(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0xEB); }
 void vporq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0xEB); }
-void vprold(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 1), x, op, T_66 | T_0F | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x72, imm); }
-void vprolq(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 1), x, op, T_66 | T_0F | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x72, imm); }
+void vprold(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 1), x, op, T_66 | T_0F | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x72, imm); }
+void vprolq(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 1), x, op, T_66 | T_0F | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x72, imm); }
 void vprolvd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x15); }
 void vprolvq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x15); }
-void vprord(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 0), x, op, T_66 | T_0F | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x72, imm); }
-void vprorq(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 0), x, op, T_66 | T_0F | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x72, imm); }
+void vprord(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 0), x, op, T_66 | T_0F | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x72, imm); }
+void vprorq(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 0), x, op, T_66 | T_0F | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x72, imm); }
 void vprorvd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x14); }
 void vprorvq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x14); }
-void vpscatterdd(const Address& addr, const Xmm& x) { opGather2(x, addr, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K | T_N4, 0xA0, 0); }
-void vpscatterdq(const Address& addr, const Xmm& x) { opGather2(x, addr, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_M_K | T_N8, 0xA0, 1); }
-void vpscatterqd(const Address& addr, const Xmm& x) { opGather2(x, addr, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K | T_N4, 0xA1, 2); }
-void vpscatterqq(const Address& addr, const Xmm& x) { opGather2(x, addr, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_M_K | T_N8, 0xA1, 0); }
+void vpscatterdd(const Address& addr, const Xmm& x) { opGather2(x, addr, T_N4 | T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K | T_VSIB, 0xA0, 0); }
+void vpscatterdq(const Address& addr, const Xmm& x) { opGather2(x, addr, T_N8 | T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_M_K | T_VSIB, 0xA0, 1); }
+void vpscatterqd(const Address& addr, const Xmm& x) { opGather2(x, addr, T_N4 | T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K | T_VSIB, 0xA1, 2); }
+void vpscatterqq(const Address& addr, const Xmm& x) { opGather2(x, addr, T_N8 | T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_M_K | T_VSIB, 0xA1, 0); }
+void vpshldd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x71, imm); }
+void vpshldq(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x71, imm); }
+void vpshldvd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x71); }
+void vpshldvq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x71); }
+void vpshldvw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX, 0x70); }
+void vpshldw(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX, 0x70, imm); }
+void vpshrdd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x73, imm); }
+void vpshrdq(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x73, imm); }
+void vpshrdvd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x73); }
+void vpshrdvq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x73); }
+void vpshrdvw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX, 0x72); }
+void vpshrdw(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX, 0x72, imm); }
+void vpshufbitqmb(const Opmask& k, const Xmm& x, const Operand& op) { opVex(k, &x, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX, 0x8F); }
 void vpsllvw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX, 0x12); }
-void vpsraq(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(Xmm(x.getKind(), 4), x, op, T_66 | T_0F | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x72, imm); }
-void vpsraq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW1 | T_YMM | T_MUST_EVEX | T_N16, 0xE2); }
+void vpsraq(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_X_XM(Xmm(x.getKind(), 4), x, op, T_66 | T_0F | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x72, imm); }
+void vpsraq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N16 | T_66 | T_0F | T_EW1 | T_YMM | T_MUST_EVEX, 0xE2); }
 void vpsravq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x46); }
 void vpsravw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX, 0x11); }
 void vpsrlvw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX, 0x10); }
-void vpternlogd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x25, imm); }
-void vpternlogq(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x25, imm); }
+void vpternlogd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x25, imm); }
+void vpternlogq(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x25, imm); }
 void vptestmb(const Opmask& k, const Xmm& x, const Operand& op) { opAVX_K_X_XM(k, x, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX, 0x26); }
 void vptestmd(const Opmask& k, const Xmm& x, const Operand& op) { opAVX_K_X_XM(k, x, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x27); }
 void vptestmq(const Opmask& k, const Xmm& x, const Operand& op) { opAVX_K_X_XM(k, x, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x27); }
@@ -1852,54 +2286,69 @@ void vptestnmq(const Opmask& k, const Xmm& x, const Operand& op) { opAVX_K_X_XM(
 void vptestnmw(const Opmask& k, const Xmm& x, const Operand& op) { opAVX_K_X_XM(k, x, op, T_F3 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX, 0x26); }
 void vpxord(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0xEF); }
 void vpxorq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0xEF); }
-void vrangepd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x50, imm); }
-void vrangeps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x50, imm); }
-void vrangesd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_SAE_X | T_MUST_EVEX | T_N8, 0x51, imm); }
-void vrangess(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_SAE_X | T_MUST_EVEX | T_N4, 0x51, imm); }
+void vrangepd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x50, imm); }
+void vrangeps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x50, imm); }
+void vrangesd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F3A | T_EW1 | T_SAE_X | T_MUST_EVEX, 0x51, imm); }
+void vrangess(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F3A | T_EW0 | T_SAE_X | T_MUST_EVEX, 0x51, imm); }
 void vrcp14pd(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x4C); }
 void vrcp14ps(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x4C); }
-void vrcp14sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_N8, 0x4D); }
-void vrcp14ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_N4, 0x4D); }
+void vrcp14sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_EW1 | T_MUST_EVEX, 0x4D); }
+void vrcp14ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_EW0 | T_MUST_EVEX, 0x4D); }
 void vrcp28pd(const Zmm& z, const Operand& op) { opAVX_X_XM_IMM(z, op, T_66 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW1 | T_B64 | T_SAE_Z, 0xCA); }
 void vrcp28ps(const Zmm& z, const Operand& op) { opAVX_X_XM_IMM(z, op, T_66 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW0 | T_B32 | T_SAE_Z, 0xCA); }
-void vrcp28sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_SAE_X | T_MUST_EVEX | T_N8, 0xCB); }
-void vrcp28ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_SAE_X | T_MUST_EVEX | T_N4, 0xCB); }
-void vreducepd(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(x, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x56, imm); }
-void vreduceps(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(x, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x56, imm); }
-void vreducesd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_SAE_X | T_MUST_EVEX | T_N8, 0x57, imm); }
-void vreducess(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_SAE_X | T_MUST_EVEX | T_N4, 0x57, imm); }
-void vrndscalepd(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(x, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x09, imm); }
-void vrndscaleps(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_XM_IMM(x, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x08, imm); }
-void vrndscalesd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW1 | T_MUST_EVEX | T_N8, 0x0B, imm); }
-void vrndscaless(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_EW0 | T_MUST_EVEX | T_N4, 0x0A, imm); }
+void vrcp28sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_EW1 | T_SAE_X | T_MUST_EVEX, 0xCB); }
+void vrcp28ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_EW0 | T_SAE_X | T_MUST_EVEX, 0xCB); }
+void vrcpph(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_MUST_EVEX | T_B16, 0x4C); }
+void vrcpsh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_MUST_EVEX, 0x4D); }
+void vreducepd(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(x, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x56, imm); }
+void vreduceph(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(x, op, T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B16, 0x56, imm); }
+void vreduceps(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(x, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x56, imm); }
+void vreducesd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F3A | T_EW1 | T_SAE_X | T_MUST_EVEX, 0x57, imm); }
+void vreducesh(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_0F3A | T_EW0 | T_SAE_X | T_MUST_EVEX, 0x57, imm); }
+void vreducess(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F3A | T_EW0 | T_SAE_X | T_MUST_EVEX, 0x57, imm); }
+void vrndscalepd(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(x, op, T_66 | T_0F3A | T_EW1 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B64, 0x09, imm); }
+void vrndscaleph(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(x, op, T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B16, 0x08, imm); }
+void vrndscaleps(const Xmm& x, const Operand& op, uint8_t imm) { opAVX_X_XM_IMM(x, op, T_66 | T_0F3A | T_EW0 | T_YMM | T_SAE_Z | T_MUST_EVEX | T_B32, 0x08, imm); }
+void vrndscalesd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F3A | T_EW1 | T_SAE_X | T_MUST_EVEX, 0x0B, imm); }
+void vrndscalesh(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_0F3A | T_EW0 | T_SAE_X | T_MUST_EVEX, 0x0A, imm); }
+void vrndscaless(const Xmm& x1, const Xmm& x2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F3A | T_EW0 | T_SAE_X | T_MUST_EVEX, 0x0A, imm); }
 void vrsqrt14pd(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x4E); }
 void vrsqrt14ps(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_B32, 0x4E); }
-void vrsqrt14sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_N8, 0x4F); }
-void vrsqrt14ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_N4, 0x4F); }
+void vrsqrt14sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX, 0x4F); }
+void vrsqrt14ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX, 0x4F); }
 void vrsqrt28pd(const Zmm& z, const Operand& op) { opAVX_X_XM_IMM(z, op, T_66 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW1 | T_B64 | T_SAE_Z, 0xCC); }
 void vrsqrt28ps(const Zmm& z, const Operand& op) { opAVX_X_XM_IMM(z, op, T_66 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW0 | T_B32 | T_SAE_Z, 0xCC); }
-void vrsqrt28sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_SAE_X | T_MUST_EVEX | T_N8, 0xCD); }
-void vrsqrt28ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_SAE_X | T_MUST_EVEX | T_N4, 0xCD); }
+void vrsqrt28sd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_EW1 | T_SAE_X | T_MUST_EVEX, 0xCD); }
+void vrsqrt28ss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_EW0 | T_SAE_X | T_MUST_EVEX, 0xCD); }
+void vrsqrtph(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_MUST_EVEX | T_B16, 0x4E); }
+void vrsqrtsh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_MUST_EVEX, 0x4F); }
 void vscalefpd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B64, 0x2C); }
+void vscalefph(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_MAP6 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0x2C); }
 void vscalefps(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B32, 0x2C); }
-void vscalefsd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW1 | T_ER_X | T_MUST_EVEX | T_N8, 0x2D); }
-void vscalefss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F38 | T_EW0 | T_ER_X | T_MUST_EVEX | T_N4, 0x2D); }
-void vscatterdpd(const Address& addr, const Xmm& x) { opGather2(x, addr, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_M_K | T_N8, 0xA2, 1); }
-void vscatterdps(const Address& addr, const Xmm& x) { opGather2(x, addr, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K | T_N4, 0xA2, 0); }
-void vscatterpf0dpd(const Address& addr) { opGatherFetch(addr, zm5, T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_N8 | T_M_K, 0xC6, Operand::YMM); }
-void vscatterpf0dps(const Address& addr) { opGatherFetch(addr, zm5, T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_N4 | T_M_K, 0xC6, Operand::ZMM); }
-void vscatterpf0qpd(const Address& addr) { opGatherFetch(addr, zm5, T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_N8 | T_M_K, 0xC7, Operand::ZMM); }
-void vscatterpf0qps(const Address& addr) { opGatherFetch(addr, zm5, T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_N4 | T_M_K, 0xC7, Operand::ZMM); }
-void vscatterpf1dpd(const Address& addr) { opGatherFetch(addr, zm6, T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_N8 | T_M_K, 0xC6, Operand::YMM); }
-void vscatterpf1dps(const Address& addr) { opGatherFetch(addr, zm6, T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_N4 | T_M_K, 0xC6, Operand::ZMM); }
-void vscatterpf1qpd(const Address& addr) { opGatherFetch(addr, zm6, T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_N8 | T_M_K, 0xC7, Operand::ZMM); }
-void vscatterpf1qps(const Address& addr) { opGatherFetch(addr, zm6, T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_N4 | T_M_K, 0xC7, Operand::ZMM); }
-void vscatterqpd(const Address& addr, const Xmm& x) { opGather2(x, addr, T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_M_K | T_N8, 0xA3, 0); }
-void vscatterqps(const Address& addr, const Xmm& x) { opGather2(x, addr, T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K | T_N4, 0xA3, 2); }
-void vshuff32x4(const Ymm& y1, const Ymm& y2, const Operand& op, uint8 imm) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F3A | T_YMM | T_MUST_EVEX | T_EW0 | T_B32, 0x23, imm); }
-void vshuff64x2(const Ymm& y1, const Ymm& y2, const Operand& op, uint8 imm) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F3A | T_YMM | T_MUST_EVEX | T_EW1 | T_B64, 0x23, imm); }
-void vshufi32x4(const Ymm& y1, const Ymm& y2, const Operand& op, uint8 imm) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F3A | T_YMM | T_MUST_EVEX | T_EW0 | T_B32, 0x43, imm); }
-void vshufi64x2(const Ymm& y1, const Ymm& y2, const Operand& op, uint8 imm) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F3A | T_YMM | T_MUST_EVEX | T_EW1 | T_B64, 0x43, imm); }
+void vscalefsd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N8 | T_66 | T_0F38 | T_EW1 | T_ER_X | T_MUST_EVEX, 0x2D); }
+void vscalefsh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_66 | T_MAP6 | T_EW0 | T_ER_X | T_MUST_EVEX, 0x2D); }
+void vscalefss(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F38 | T_EW0 | T_ER_X | T_MUST_EVEX, 0x2D); }
+void vscatterdpd(const Address& addr, const Xmm& x) { opGather2(x, addr, T_N8 | T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_M_K | T_VSIB, 0xA2, 1); }
+void vscatterdps(const Address& addr, const Xmm& x) { opGather2(x, addr, T_N4 | T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K | T_VSIB, 0xA2, 0); }
+void vscatterpf0dpd(const Address& addr) { opGatherFetch(addr, zm5, T_N8 | T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC6, Operand::YMM); }
+void vscatterpf0dps(const Address& addr) { opGatherFetch(addr, zm5, T_N4 | T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC6, Operand::ZMM); }
+void vscatterpf0qpd(const Address& addr) { opGatherFetch(addr, zm5, T_N8 | T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC7, Operand::ZMM); }
+void vscatterpf0qps(const Address& addr) { opGatherFetch(addr, zm5, T_N4 | T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC7, Operand::ZMM); }
+void vscatterpf1dpd(const Address& addr) { opGatherFetch(addr, zm6, T_N8 | T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC6, Operand::YMM); }
+void vscatterpf1dps(const Address& addr) { opGatherFetch(addr, zm6, T_N4 | T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC6, Operand::ZMM); }
+void vscatterpf1qpd(const Address& addr) { opGatherFetch(addr, zm6, T_N8 | T_66 | T_0F38 | T_EW1 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC7, Operand::ZMM); }
+void vscatterpf1qps(const Address& addr) { opGatherFetch(addr, zm6, T_N4 | T_66 | T_0F38 | T_EW0 | T_MUST_EVEX | T_M_K | T_VSIB, 0xC7, Operand::ZMM); }
+void vscatterqpd(const Address& addr, const Xmm& x) { opGather2(x, addr, T_N8 | T_66 | T_0F38 | T_EW1 | T_YMM | T_MUST_EVEX | T_M_K | T_VSIB, 0xA3, 0); }
+void vscatterqps(const Address& addr, const Xmm& x) { opGather2(x, addr, T_N4 | T_66 | T_0F38 | T_EW0 | T_YMM | T_MUST_EVEX | T_M_K | T_VSIB, 0xA3, 2); }
+void vshuff32x4(const Ymm& y1, const Ymm& y2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F3A | T_YMM | T_MUST_EVEX | T_EW0 | T_B32, 0x23, imm); }
+void vshuff64x2(const Ymm& y1, const Ymm& y2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F3A | T_YMM | T_MUST_EVEX | T_EW1 | T_B64, 0x23, imm); }
+void vshufi32x4(const Ymm& y1, const Ymm& y2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F3A | T_YMM | T_MUST_EVEX | T_EW0 | T_B32, 0x43, imm); }
+void vshufi64x2(const Ymm& y1, const Ymm& y2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F3A | T_YMM | T_MUST_EVEX | T_EW1 | T_B64, 0x43, imm); }
+void vsqrtph(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_MAP5 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0x51); }
+void vsqrtsh(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_N2 | T_F3 | T_MAP5 | T_EW0 | T_ER_X | T_MUST_EVEX, 0x51); }
+void vsubph(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_EW0 | T_YMM | T_MUST_EVEX | T_ER_Z | T_B16, 0x5C); }
+void vsubsh(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_F3 | T_EW0 | T_MUST_EVEX | T_ER_X | T_N2, 0x5C); }
+void vucomish(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_MAP5 | T_MUST_EVEX | T_EW0 | T_SAE_X | T_N2, 0x2E); }
 #ifdef XBYAK64
 void kmovq(const Opmask& k, const Reg64& r) { opVex(k, 0, r, T_L0 | T_0F | T_F2 | T_W1, 0x92); }
 void kmovq(const Reg64& r, const Opmask& k) { opVex(r, 0, k, T_L0 | T_0F | T_F2 | T_W1, 0x93); }
diff --git a/dep/xbyak/xbyak/xbyak_util.h b/dep/xbyak/xbyak/xbyak_util.h
index 5ba62676e..0bcc13499 100644
--- a/dep/xbyak/xbyak/xbyak_util.h
+++ b/dep/xbyak/xbyak/xbyak_util.h
@@ -1,16 +1,39 @@
 #ifndef XBYAK_XBYAK_UTIL_H_
 #define XBYAK_XBYAK_UTIL_H_
 
+#ifdef XBYAK_ONLY_CLASS_CPU
+#include <stdint.h>
+#include <stdlib.h>
+#include <assert.h>
+#ifndef XBYAK_THROW
+	#define XBYAK_THROW(x) ;
+	#define XBYAK_THROW_RET(x, y) return y;
+#endif
+#ifndef XBYAK_CONSTEXPR
+#if ((__cplusplus >= 201402L) && !(!defined(__clang__) && defined(__GNUC__) && (__GNUC__ <= 5))) || (defined(_MSC_VER) && _MSC_VER >= 1910)
+	#define XBYAK_CONSTEXPR constexpr
+#else
+	#define XBYAK_CONSTEXPR
+#endif
+#endif
+#else
+#include <string.h>
+
 /**
 	utility class and functions for Xbyak
 	Xbyak::util::Clock ; rdtsc timer
 	Xbyak::util::Cpu ; detect CPU
-	@note this header is UNDER CONSTRUCTION!
 */
 #include "xbyak.h"
+#endif // XBYAK_ONLY_CLASS_CPU
+
+#if defined(__i386__) || defined(__x86_64__) || defined(_M_IX86) || defined(_M_X64)
+	#define XBYAK_INTEL_CPU_SPECIFIC
+#endif
 
-#ifdef _MSC_VER
-	#if (_MSC_VER < 1400) && defined(XBYAK32)
+#ifdef XBYAK_INTEL_CPU_SPECIFIC
+#ifdef _WIN32
+	#if defined(_MSC_VER) && (_MSC_VER < 1400) && defined(XBYAK32)
 		static inline __declspec(naked) void __cpuid(int[4], int)
 		{
 			__asm {
@@ -47,25 +70,92 @@
 		#endif
 	#endif
 #endif
+#endif
+
+#ifdef XBYAK_USE_VTUNE
+	// -I /opt/intel/vtune_amplifier/include/ -L /opt/intel/vtune_amplifier/lib64 -ljitprofiling -ldl
+	#include <jitprofiling.h>
+	#ifdef _MSC_VER
+		#pragma comment(lib, "libittnotify.lib")
+	#endif
+	#ifdef __linux__
+		#include <dlfcn.h>
+	#endif
+#endif
+#ifdef __linux__
+	#define XBYAK_USE_PERF
+#endif
 
 namespace Xbyak { namespace util {
 
+typedef enum {
+   SmtLevel = 1,
+   CoreLevel = 2
+} IntelCpuTopologyLevel;
+
+namespace local {
+
+template<uint64_t L, uint64_t H = 0>
+struct TypeT {
+};
+
+template<uint64_t L1, uint64_t H1, uint64_t L2, uint64_t H2>
+XBYAK_CONSTEXPR TypeT<L1 | L2, H1 | H2> operator|(TypeT<L1, H1>, TypeT<L2, H2>) { return TypeT<L1 | L2, H1 | H2>(); }
+
+template<typename T>
+inline T max_(T x, T y) { return x >= y ? x : y; }
+template<typename T>
+inline T min_(T x, T y) { return x < y ? x : y; }
+
+} // local
+
 /**
 	CPU detection class
+	@note static inline const member is supported by c++17 or later, so use template hack
 */
 class Cpu {
-	uint64 type_;
-	unsigned int get32bitAsBE(const char *x) const
+public:
+	class Type {
+		uint64_t L;
+		uint64_t H;
+	public:
+		Type(uint64_t L = 0, uint64_t H = 0) : L(L), H(H) { }
+		template<uint64_t L_, uint64_t H_>
+		Type(local::TypeT<L_, H_>) : L(L_), H(H_) {}
+		Type& operator&=(const Type& rhs) { L &= rhs.L; H &= rhs.H; return *this; }
+		Type& operator|=(const Type& rhs) { L |= rhs.L; H |= rhs.H; return *this; }
+		Type operator&(const Type& rhs) const { Type t = *this; t &= rhs; return t; }
+		Type operator|(const Type& rhs) const { Type t = *this; t |= rhs; return t; }
+		bool operator==(const Type& rhs) const { return H == rhs.H && L == rhs.L; }
+		bool operator!=(const Type& rhs) const { return !operator==(rhs); }
+		// without explicit because backward compatilibity
+		operator bool() const { return (H | L) != 0; }
+		uint64_t getL() const { return L; }
+		uint64_t getH() const { return H; }
+	};
+private:
+	Type type_;
+	//system topology
+	bool x2APIC_supported_;
+	static const size_t maxTopologyLevels = 2;
+	uint32_t numCores_[maxTopologyLevels];
+
+	static const uint32_t maxNumberCacheLevels = 10;
+	uint32_t dataCacheSize_[maxNumberCacheLevels];
+	uint32_t coresSharignDataCache_[maxNumberCacheLevels];
+	uint32_t dataCacheLevels_;
+
+	uint32_t get32bitAsBE(const char *x) const
 	{
 		return x[0] | (x[1] << 8) | (x[2] << 16) | (x[3] << 24);
 	}
-	unsigned int mask(int n) const
+	uint32_t mask(int n) const
 	{
 		return (1U << n) - 1;
 	}
 	void setFamily()
 	{
-		unsigned int data[4];
+		uint32_t data[4] = {};
 		getCpuid(1, data);
 		stepping = data[0] & mask(4);
 		model = (data[0] >> 4) & mask(4);
@@ -84,6 +174,135 @@ class Cpu {
 			displayModel = model;
 		}
 	}
+	uint32_t extractBit(uint32_t val, uint32_t base, uint32_t end)
+	{
+		return (val >> base) & ((1u << (end - base)) - 1);
+	}
+	void setNumCores()
+	{
+		if (!has(tINTEL) && !has(tAMD)) return;
+
+		uint32_t data[4] = {};
+		getCpuidEx(0x0, 0, data);
+		if (data[0] >= 0xB) {
+			 /*
+				if leaf 11 exists(x2APIC is supported),
+				we use it to get the number of smt cores and cores on socket
+
+				leaf 0xB can be zeroed-out by a hypervisor
+			*/
+			x2APIC_supported_ = true;
+			for (uint32_t i = 0; i < maxTopologyLevels; i++) {
+				getCpuidEx(0xB, i, data);
+				IntelCpuTopologyLevel level = (IntelCpuTopologyLevel)extractBit(data[2], 8, 15);
+				if (level == SmtLevel || level == CoreLevel) {
+					numCores_[level - 1] = extractBit(data[1], 0, 15);
+				}
+			}
+			/*
+				Fallback values in case a hypervisor has 0xB leaf zeroed-out.
+			*/
+			numCores_[SmtLevel - 1] = local::max_(1u, numCores_[SmtLevel - 1]);
+			numCores_[CoreLevel - 1] = local::max_(numCores_[SmtLevel - 1], numCores_[CoreLevel - 1]);
+		} else {
+			/*
+				Failed to deremine num of cores without x2APIC support.
+				TODO: USE initial APIC ID to determine ncores.
+			*/
+			numCores_[SmtLevel - 1] = 0;
+			numCores_[CoreLevel - 1] = 0;
+		}
+
+	}
+	void setCacheHierarchy()
+	{
+		if (!has(tINTEL) && !has(tAMD)) return;
+
+		// https://github.com/amd/ZenDNN/blob/a08bf9a9efc160a69147cdecfb61cc85cc0d4928/src/cpu/x64/xbyak/xbyak_util.h#L236-L288
+		if (has(tAMD)) {
+			// There are 3 Data Cache Levels (L1, L2, L3)
+			dataCacheLevels_ = 3;
+			const uint32_t leaf = 0x8000001D; // for modern AMD CPus
+			// Sub leaf value ranges from 0 to 3
+			// Sub leaf value 0 refers to L1 Data Cache
+			// Sub leaf value 1 refers to L1 Instruction Cache
+			// Sub leaf value 2 refers to L2 Cache
+			// Sub leaf value 3 refers to L3 Cache
+			// For legacy AMD CPU, use leaf 0x80000005 for L1 cache
+			// and 0x80000006 for L2 and L3 cache
+			int cache_index = 0;
+			for (uint32_t sub_leaf = 0; sub_leaf <= dataCacheLevels_; sub_leaf++) {
+				// Skip sub_leaf = 1 as it refers to
+				// L1 Instruction Cache (not required)
+				if (sub_leaf == 1) {
+					continue;
+				}
+				uint32_t data[4] = {};
+				getCpuidEx(leaf, sub_leaf, data);
+				// Cache Size = Line Size * Partitions * Associativity * Cache Sets
+				dataCacheSize_[cache_index] =
+					(extractBit(data[1], 22, 31) + 1) // Associativity-1
+					* (extractBit(data[1], 12, 21) + 1) // Partitions-1
+					* (extractBit(data[1], 0, 11) + 1) // Line Size
+					* (data[2] + 1);
+				// Calculate the number of cores sharing the current data cache
+				int smt_width = numCores_[0];
+				int logical_cores = numCores_[1];
+				int actual_logical_cores = extractBit(data[0], 14, 25) /* # of cores * # of threads */ + 1;
+				if (logical_cores != 0) {
+					actual_logical_cores = local::min_(actual_logical_cores, logical_cores);
+				}
+				coresSharignDataCache_[cache_index] = local::max_(actual_logical_cores / smt_width, 1);
+				++cache_index;
+			}
+			return;
+		}
+		// intel
+		const uint32_t NO_CACHE = 0;
+		const uint32_t DATA_CACHE = 1;
+//		const uint32_t INSTRUCTION_CACHE = 2;
+		const uint32_t UNIFIED_CACHE = 3;
+		uint32_t smt_width = 0;
+		uint32_t logical_cores = 0;
+		uint32_t data[4] = {};
+
+		if (x2APIC_supported_) {
+			smt_width = numCores_[0];
+			logical_cores = numCores_[1];
+		}
+
+		/*
+			Assumptions:
+			the first level of data cache is not shared (which is the
+			case for every existing architecture) and use this to
+			determine the SMT width for arch not supporting leaf 11.
+			when leaf 4 reports a number of core less than numCores_
+			on socket reported by leaf 11, then it is a correct number
+			of cores not an upperbound.
+		*/
+		for (int i = 0; dataCacheLevels_ < maxNumberCacheLevels; i++) {
+			getCpuidEx(0x4, i, data);
+			uint32_t cacheType = extractBit(data[0], 0, 4);
+			if (cacheType == NO_CACHE) break;
+			if (cacheType == DATA_CACHE || cacheType == UNIFIED_CACHE) {
+				uint32_t actual_logical_cores = extractBit(data[0], 14, 25) + 1;
+				if (logical_cores != 0) { // true only if leaf 0xB is supported and valid
+					actual_logical_cores = local::min_(actual_logical_cores, logical_cores);
+				}
+				assert(actual_logical_cores != 0);
+				dataCacheSize_[dataCacheLevels_] =
+					(extractBit(data[1], 22, 31) + 1)
+					* (extractBit(data[1], 12, 21) + 1)
+					* (extractBit(data[1], 0, 11) + 1)
+					* (data[2] + 1);
+				if (cacheType == DATA_CACHE && smt_width == 0) smt_width = actual_logical_cores;
+				assert(smt_width != 0);
+				coresSharignDataCache_[dataCacheLevels_] = local::max_(actual_logical_cores / smt_width, 1u);
+				dataCacheLevels_++;
+			}
+		}
+	}
+
 public:
 	int model;
 	int family;
@@ -92,192 +311,358 @@ public:
 	int extFamily;
 	int displayFamily; // family + extFamily
 	int displayModel; // model + extModel
+
+	uint32_t getNumCores(IntelCpuTopologyLevel level) const {
+		if (!x2APIC_supported_) XBYAK_THROW_RET(ERR_X2APIC_IS_NOT_SUPPORTED, 0)
+		switch (level) {
+		case SmtLevel: return numCores_[level - 1];
+		case CoreLevel: return numCores_[level - 1] / numCores_[SmtLevel - 1];
+		default: XBYAK_THROW_RET(ERR_X2APIC_IS_NOT_SUPPORTED, 0)
+		}
+	}
+
+	uint32_t getDataCacheLevels() const { return dataCacheLevels_; }
+	uint32_t getCoresSharingDataCache(uint32_t i) const
+	{
+		if (i >= dataCacheLevels_) XBYAK_THROW_RET(ERR_BAD_PARAMETER, 0)
+		return coresSharignDataCache_[i];
+	}
+	uint32_t getDataCacheSize(uint32_t i) const
+	{
+		if (i >= dataCacheLevels_) XBYAK_THROW_RET(ERR_BAD_PARAMETER, 0)
+		return dataCacheSize_[i];
+	}
+
 	/*
 		data[] = { eax, ebx, ecx, edx }
 	*/
-	static inline void getCpuid(unsigned int eaxIn, unsigned int data[4])
+	static inline void getCpuid(uint32_t eaxIn, uint32_t data[4])
 	{
-#ifdef _MSC_VER
+#ifdef XBYAK_INTEL_CPU_SPECIFIC
+	#ifdef _WIN32
 		__cpuid(reinterpret_cast<int*>(data), eaxIn);
-#else
+	#else
 		__cpuid(eaxIn, data[0], data[1], data[2], data[3]);
+	#endif
+#else
+		(void)eaxIn;
+		(void)data;
 #endif
 	}
-	static inline void getCpuidEx(unsigned int eaxIn, unsigned int ecxIn, unsigned int data[4])
+	static inline void getCpuidEx(uint32_t eaxIn, uint32_t ecxIn, uint32_t data[4])
 	{
-#ifdef _MSC_VER
+#ifdef XBYAK_INTEL_CPU_SPECIFIC
+	#ifdef _WIN32
 		__cpuidex(reinterpret_cast<int*>(data), eaxIn, ecxIn);
-#else
+	#else
 		__cpuid_count(eaxIn, ecxIn, data[0], data[1], data[2], data[3]);
+	#endif
+#else
+		(void)eaxIn;
+		(void)ecxIn;
+		(void)data;
 #endif
 	}
-	static inline uint64 getXfeature()
+	static inline uint64_t getXfeature()
 	{
-#ifdef _MSC_VER
+#ifdef XBYAK_INTEL_CPU_SPECIFIC
+	#ifdef _MSC_VER
 		return _xgetbv(0);
-#else
-		unsigned int eax, edx;
+	#else
+		uint32_t eax, edx;
 		// xgetvb is not support on gcc 4.2
 //		__asm__ volatile("xgetbv" : "=a"(eax), "=d"(edx) : "c"(0));
 		__asm__ volatile(".byte 0x0f, 0x01, 0xd0" : "=a"(eax), "=d"(edx) : "c"(0));
-		return ((uint64)edx << 32) | eax;
+		return ((uint64_t)edx << 32) | eax;
+	#endif
+#else
+		return 0;
 #endif
 	}
-	typedef uint64 Type;
-	static const Type NONE = 0;
-	static const Type tMMX = 1 << 0;
-	static const Type tMMX2 = 1 << 1;
-	static const Type tCMOV = 1 << 2;
-	static const Type tSSE = 1 << 3;
-	static const Type tSSE2 = 1 << 4;
-	static const Type tSSE3 = 1 << 5;
-	static const Type tSSSE3 = 1 << 6;
-	static const Type tSSE41 = 1 << 7;
-	static const Type tSSE42 = 1 << 8;
-	static const Type tPOPCNT = 1 << 9;
-	static const Type tAESNI = 1 << 10;
-	static const Type tSSE5 = 1 << 11;
-	static const Type tOSXSAVE = 1 << 12;
-	static const Type tPCLMULQDQ = 1 << 13;
-	static const Type tAVX = 1 << 14;
-	static const Type tFMA = 1 << 15;
-
-	static const Type t3DN = 1 << 16;
-	static const Type tE3DN = 1 << 17;
-	static const Type tSSE4a = 1 << 18;
-	static const Type tRDTSCP = 1 << 19;
-	static const Type tAVX2 = 1 << 20;
-	static const Type tBMI1 = 1 << 21; // andn, bextr, blsi, blsmsk, blsr, tzcnt
-	static const Type tBMI2 = 1 << 22; // bzhi, mulx, pdep, pext, rorx, sarx, shlx, shrx
-	static const Type tLZCNT = 1 << 23;
-
-	static const Type tINTEL = 1 << 24;
-	static const Type tAMD = 1 << 25;
-
-	static const Type tENHANCED_REP = 1 << 26; // enhanced rep movsb/stosb
-	static const Type tRDRAND = 1 << 27;
-	static const Type tADX = 1 << 28; // adcx, adox
-	static const Type tRDSEED = 1 << 29; // rdseed
-	static const Type tSMAP = 1 << 30; // stac
-	static const Type tHLE = uint64(1) << 31; // xacquire, xrelease, xtest
-	static const Type tRTM = uint64(1) << 32; // xbegin, xend, xabort
-	static const Type tF16C = uint64(1) << 33; // vcvtph2ps, vcvtps2ph
-	static const Type tMOVBE = uint64(1) << 34; // mobve
-	static const Type tAVX512F = uint64(1) << 35;
-	static const Type tAVX512DQ = uint64(1) << 36;
-	static const Type tAVX512IFMA = uint64(1) << 37;
-	static const Type tAVX512PF = uint64(1) << 38;
-	static const Type tAVX512ER = uint64(1) << 39;
-	static const Type tAVX512CD = uint64(1) << 40;
-	static const Type tAVX512BW = uint64(1) << 41;
-	static const Type tAVX512VL = uint64(1) << 42;
-	static const Type tAVX512VBMI = uint64(1) << 43;
-	static const Type tAVX512_4VNNIW = uint64(1) << 44;
-	static const Type tAVX512_4FMAPS = uint64(1) << 45;
-	static const Type tPREFETCHWT1 = uint64(1) << 46;
+
+#define XBYAK_SPLIT_ID(id) ((0 <= id && id < 64) ? (1ull << (id % 64)) : 0), (id >= 64 ? (1ull << (id % 64)) : 0)
+#if (__cplusplus >= 201103) || (defined(_MSC_VER) && (_MSC_VER >= 1700)) /* VS2012 */
+	#define XBYAK_DEFINE_TYPE(id, NAME) static const constexpr local::TypeT<XBYAK_SPLIT_ID(id)> NAME{}
+#else
+	#define XBYAK_DEFINE_TYPE(id, NAME) static const local::TypeT<XBYAK_SPLIT_ID(id)> NAME
+#endif
+	XBYAK_DEFINE_TYPE(0, tMMX);
+	XBYAK_DEFINE_TYPE(1, tMMX2);
+	XBYAK_DEFINE_TYPE(2, tCMOV);
+	XBYAK_DEFINE_TYPE(3, tSSE);
+	XBYAK_DEFINE_TYPE(4, tSSE2);
+	XBYAK_DEFINE_TYPE(5, tSSE3);
+	XBYAK_DEFINE_TYPE(6, tSSSE3);
+	XBYAK_DEFINE_TYPE(7, tSSE41);
+	XBYAK_DEFINE_TYPE(8, tSSE42);
+	XBYAK_DEFINE_TYPE(9, tPOPCNT);
+	XBYAK_DEFINE_TYPE(10, tAESNI);
+	XBYAK_DEFINE_TYPE(11, tAVX512_FP16);
+	XBYAK_DEFINE_TYPE(12, tOSXSAVE);
+	XBYAK_DEFINE_TYPE(13, tPCLMULQDQ);
+	XBYAK_DEFINE_TYPE(14, tAVX);
+	XBYAK_DEFINE_TYPE(15, tFMA);
+	XBYAK_DEFINE_TYPE(16, t3DN);
+	XBYAK_DEFINE_TYPE(17, tE3DN);
+	XBYAK_DEFINE_TYPE(18, tWAITPKG);
+	XBYAK_DEFINE_TYPE(19, tRDTSCP);
+	XBYAK_DEFINE_TYPE(20, tAVX2);
+	XBYAK_DEFINE_TYPE(21, tBMI1); // andn, bextr, blsi, blsmsk, blsr, tzcnt
+	XBYAK_DEFINE_TYPE(22, tBMI2); // bzhi, mulx, pdep, pext, rorx, sarx, shlx, shrx
+	XBYAK_DEFINE_TYPE(23, tLZCNT);
+	XBYAK_DEFINE_TYPE(24, tINTEL);
+	XBYAK_DEFINE_TYPE(25, tAMD);
+	XBYAK_DEFINE_TYPE(26, tENHANCED_REP); // enhanced rep movsb/stosb
+	XBYAK_DEFINE_TYPE(27, tRDRAND);
+	XBYAK_DEFINE_TYPE(28, tADX); // adcx, adox
+	XBYAK_DEFINE_TYPE(29, tRDSEED); // rdseed
+	XBYAK_DEFINE_TYPE(30, tSMAP); // stac
+	XBYAK_DEFINE_TYPE(31, tHLE); // xacquire, xrelease, xtest
+	XBYAK_DEFINE_TYPE(32, tRTM); // xbegin, xend, xabort
+	XBYAK_DEFINE_TYPE(33, tF16C); // vcvtph2ps, vcvtps2ph
+	XBYAK_DEFINE_TYPE(34, tMOVBE); // mobve
+	XBYAK_DEFINE_TYPE(35, tAVX512F);
+	XBYAK_DEFINE_TYPE(36, tAVX512DQ);
+	XBYAK_DEFINE_TYPE(37, tAVX512_IFMA);
+	XBYAK_DEFINE_TYPE(37, tAVX512IFMA);// = tAVX512_IFMA;
+	XBYAK_DEFINE_TYPE(38, tAVX512PF);
+	XBYAK_DEFINE_TYPE(39, tAVX512ER);
+	XBYAK_DEFINE_TYPE(40, tAVX512CD);
+	XBYAK_DEFINE_TYPE(41, tAVX512BW);
+	XBYAK_DEFINE_TYPE(42, tAVX512VL);
+	XBYAK_DEFINE_TYPE(43, tAVX512_VBMI);
+	XBYAK_DEFINE_TYPE(43, tAVX512VBMI); // = tAVX512_VBMI; // changed by Intel's manual
+	XBYAK_DEFINE_TYPE(44, tAVX512_4VNNIW);
+	XBYAK_DEFINE_TYPE(45, tAVX512_4FMAPS);
+	XBYAK_DEFINE_TYPE(46, tPREFETCHWT1);
+	XBYAK_DEFINE_TYPE(47, tPREFETCHW);
+	XBYAK_DEFINE_TYPE(48, tSHA);
+	XBYAK_DEFINE_TYPE(49, tMPX);
+	XBYAK_DEFINE_TYPE(50, tAVX512_VBMI2);
+	XBYAK_DEFINE_TYPE(51, tGFNI);
+	XBYAK_DEFINE_TYPE(52, tVAES);
+	XBYAK_DEFINE_TYPE(53, tVPCLMULQDQ);
+	XBYAK_DEFINE_TYPE(54, tAVX512_VNNI);
+	XBYAK_DEFINE_TYPE(55, tAVX512_BITALG);
+	XBYAK_DEFINE_TYPE(56, tAVX512_VPOPCNTDQ);
+	XBYAK_DEFINE_TYPE(57, tAVX512_BF16);
+	XBYAK_DEFINE_TYPE(58, tAVX512_VP2INTERSECT);
+	XBYAK_DEFINE_TYPE(59, tAMX_TILE);
+	XBYAK_DEFINE_TYPE(60, tAMX_INT8);
+	XBYAK_DEFINE_TYPE(61, tAMX_BF16);
+	XBYAK_DEFINE_TYPE(62, tAVX_VNNI);
+	XBYAK_DEFINE_TYPE(63, tCLFLUSHOPT);
+	XBYAK_DEFINE_TYPE(64, tCLDEMOTE);
+	XBYAK_DEFINE_TYPE(65, tMOVDIRI);
+	XBYAK_DEFINE_TYPE(66, tMOVDIR64B);
+	XBYAK_DEFINE_TYPE(67, tCLZERO); // AMD Zen
+	XBYAK_DEFINE_TYPE(68, tAMX_FP16);
+	XBYAK_DEFINE_TYPE(69, tAVX_VNNI_INT8);
+	XBYAK_DEFINE_TYPE(70, tAVX_NE_CONVERT);
+	XBYAK_DEFINE_TYPE(71, tAVX_IFMA);
+	XBYAK_DEFINE_TYPE(72, tRAO_INT);
+	XBYAK_DEFINE_TYPE(73, tCMPCCXADD);
+	XBYAK_DEFINE_TYPE(74, tPREFETCHITI);
+	XBYAK_DEFINE_TYPE(75, tSERIALIZE);
+	XBYAK_DEFINE_TYPE(76, tUINTR);
+	XBYAK_DEFINE_TYPE(77, tXSAVE);
+	XBYAK_DEFINE_TYPE(78, tSHA512);
+	XBYAK_DEFINE_TYPE(79, tSM3);
+	XBYAK_DEFINE_TYPE(80, tSM4);
+	XBYAK_DEFINE_TYPE(81, tAVX_VNNI_INT16);
+
+#undef XBYAK_SPLIT_ID
+#undef XBYAK_DEFINE_TYPE
 
 	Cpu()
-		: type_(NONE)
+		: type_()
+		, x2APIC_supported_(false)
+		, numCores_()
+		, dataCacheSize_()
+		, coresSharignDataCache_()
+		, dataCacheLevels_(0)
 	{
-		unsigned int data[4];
+		uint32_t data[4] = {};
+		const uint32_t& EAX = data[0];
+		const uint32_t& EBX = data[1];
+		const uint32_t& ECX = data[2];
+		const uint32_t& EDX = data[3];
 		getCpuid(0, data);
-		const unsigned int maxNum = data[0];
+		const uint32_t maxNum = EAX;
 		static const char intel[] = "ntel";
 		static const char amd[] = "cAMD";
-		if (data[2] == get32bitAsBE(amd)) {
+		if (ECX == get32bitAsBE(amd)) {
 			type_ |= tAMD;
 			getCpuid(0x80000001, data);
-			if (data[3] & (1U << 31)) type_ |= t3DN;
-			if (data[3] & (1U << 15)) type_ |= tCMOV;
-			if (data[3] & (1U << 30)) type_ |= tE3DN;
-			if (data[3] & (1U << 22)) type_ |= tMMX2;
-			if (data[3] & (1U << 27)) type_ |= tRDTSCP;
+			if (EDX & (1U << 31)) {
+				type_ |= t3DN;
+				// 3DNow! implies support for PREFETCHW on AMD
+				type_ |= tPREFETCHW;
+			}
+
+			if (EDX & (1U << 29)) {
+				// Long mode implies support for PREFETCHW on AMD
+				type_ |= tPREFETCHW;
+			}
 		}
-		if (data[2] == get32bitAsBE(intel)) {
+		if (ECX == get32bitAsBE(intel)) {
 			type_ |= tINTEL;
+		}
+
+		// Extended flags information
+		getCpuid(0x80000000, data);
+		const uint32_t maxExtendedNum = EAX;
+		if (maxExtendedNum >= 0x80000001) {
 			getCpuid(0x80000001, data);
-			if (data[3] & (1U << 27)) type_ |= tRDTSCP;
-			if (data[2] & (1U << 5)) type_ |= tLZCNT;
+
+			if (EDX & (1U << 31)) type_ |= t3DN;
+			if (EDX & (1U << 30)) type_ |= tE3DN;
+			if (EDX & (1U << 27)) type_ |= tRDTSCP;
+			if (EDX & (1U << 22)) type_ |= tMMX2;
+			if (EDX & (1U << 15)) type_ |= tCMOV;
+			if (ECX & (1U << 5)) type_ |= tLZCNT;
+			if (ECX & (1U << 8)) type_ |= tPREFETCHW;
 		}
+
+		if (maxExtendedNum >= 0x80000008) {
+			getCpuid(0x80000008, data);
+			if (EBX & (1U << 0)) type_ |= tCLZERO;
+		}
+
 		getCpuid(1, data);
-		if (data[2] & (1U << 0)) type_ |= tSSE3;
-		if (data[2] & (1U << 9)) type_ |= tSSSE3;
-		if (data[2] & (1U << 19)) type_ |= tSSE41;
-		if (data[2] & (1U << 20)) type_ |= tSSE42;
-		if (data[2] & (1U << 22)) type_ |= tMOVBE;
-		if (data[2] & (1U << 23)) type_ |= tPOPCNT;
-		if (data[2] & (1U << 25)) type_ |= tAESNI;
-		if (data[2] & (1U << 1)) type_ |= tPCLMULQDQ;
-		if (data[2] & (1U << 27)) type_ |= tOSXSAVE;
-		if (data[2] & (1U << 30)) type_ |= tRDRAND;
-		if (data[2] & (1U << 29)) type_ |= tF16C;
-
-		if (data[3] & (1U << 15)) type_ |= tCMOV;
-		if (data[3] & (1U << 23)) type_ |= tMMX;
-		if (data[3] & (1U << 25)) type_ |= tMMX2 | tSSE;
-		if (data[3] & (1U << 26)) type_ |= tSSE2;
+		if (ECX & (1U << 0)) type_ |= tSSE3;
+		if (ECX & (1U << 1)) type_ |= tPCLMULQDQ;
+		if (ECX & (1U << 9)) type_ |= tSSSE3;
+		if (ECX & (1U << 19)) type_ |= tSSE41;
+		if (ECX & (1U << 20)) type_ |= tSSE42;
+		if (ECX & (1U << 22)) type_ |= tMOVBE;
+		if (ECX & (1U << 23)) type_ |= tPOPCNT;
+		if (ECX & (1U << 25)) type_ |= tAESNI;
+		if (ECX & (1U << 26)) type_ |= tXSAVE;
+		if (ECX & (1U << 27)) type_ |= tOSXSAVE;
+		if (ECX & (1U << 30)) type_ |= tRDRAND;
+		if (ECX & (1U << 29)) type_ |= tF16C;
+
+		if (EDX & (1U << 15)) type_ |= tCMOV;
+		if (EDX & (1U << 23)) type_ |= tMMX;
+		if (EDX & (1U << 25)) type_ |= tMMX2 | tSSE;
+		if (EDX & (1U << 26)) type_ |= tSSE2;
 
 		if (type_ & tOSXSAVE) {
 			// check XFEATURE_ENABLED_MASK[2:1] = '11b'
-			uint64 bv = getXfeature();
+			uint64_t bv = getXfeature();
 			if ((bv & 6) == 6) {
-				if (data[2] & (1U << 28)) type_ |= tAVX;
-				if (data[2] & (1U << 12)) type_ |= tFMA;
-				if (((bv >> 5) & 7) == 7) {
+				if (ECX & (1U << 28)) type_ |= tAVX;
+				if (ECX & (1U << 12)) type_ |= tFMA;
+				// do *not* check AVX-512 state on macOS because it has on-demand AVX-512 support
+#if !defined(__APPLE__)
+				if (((bv >> 5) & 7) == 7)
+#endif
+				{
 					getCpuidEx(7, 0, data);
-					if (data[1] & (1U << 16)) type_ |= tAVX512F;
+					if (EBX & (1U << 16)) type_ |= tAVX512F;
 					if (type_ & tAVX512F) {
-						if (data[1] & (1U << 17)) type_ |= tAVX512DQ;
-						if (data[1] & (1U << 21)) type_ |= tAVX512IFMA;
-						if (data[1] & (1U << 26)) type_ |= tAVX512PF;
-						if (data[1] & (1U << 27)) type_ |= tAVX512ER;
-						if (data[1] & (1U << 28)) type_ |= tAVX512CD;
-						if (data[1] & (1U << 30)) type_ |= tAVX512BW;
-						if (data[1] & (1U << 31)) type_ |= tAVX512VL;
-						if (data[2] & (1U << 1)) type_ |= tAVX512VBMI;
-						if (data[3] & (1U << 2)) type_ |= tAVX512_4VNNIW;
-						if (data[3] & (1U << 3)) type_ |= tAVX512_4FMAPS;
+						if (EBX & (1U << 17)) type_ |= tAVX512DQ;
+						if (EBX & (1U << 21)) type_ |= tAVX512_IFMA;
+						if (EBX & (1U << 26)) type_ |= tAVX512PF;
+						if (EBX & (1U << 27)) type_ |= tAVX512ER;
+						if (EBX & (1U << 28)) type_ |= tAVX512CD;
+						if (EBX & (1U << 30)) type_ |= tAVX512BW;
+						if (EBX & (1U << 31)) type_ |= tAVX512VL;
+						if (ECX & (1U << 1)) type_ |= tAVX512_VBMI;
+						if (ECX & (1U << 6)) type_ |= tAVX512_VBMI2;
+						if (ECX & (1U << 11)) type_ |= tAVX512_VNNI;
+						if (ECX & (1U << 12)) type_ |= tAVX512_BITALG;
+						if (ECX & (1U << 14)) type_ |= tAVX512_VPOPCNTDQ;
+						if (EDX & (1U << 2)) type_ |= tAVX512_4VNNIW;
+						if (EDX & (1U << 3)) type_ |= tAVX512_4FMAPS;
+						if (EDX & (1U << 8)) type_ |= tAVX512_VP2INTERSECT;
+						if ((type_ & tAVX512BW) && (EDX & (1U << 23))) type_ |= tAVX512_FP16;
 					}
 				}
 			}
 		}
 		if (maxNum >= 7) {
 			getCpuidEx(7, 0, data);
-			if (type_ & tAVX && data[1] & 0x20) type_ |= tAVX2;
-			if (data[1] & (1U << 3)) type_ |= tBMI1;
-			if (data[1] & (1U << 8)) type_ |= tBMI2;
-			if (data[1] & (1U << 9)) type_ |= tENHANCED_REP;
-			if (data[1] & (1U << 18)) type_ |= tRDSEED;
-			if (data[1] & (1U << 19)) type_ |= tADX;
-			if (data[1] & (1U << 20)) type_ |= tSMAP;
-			if (data[1] & (1U << 4)) type_ |= tHLE;
-			if (data[1] & (1U << 11)) type_ |= tRTM;
-			if (data[2] & (1U << 0)) type_ |= tPREFETCHWT1;
+			const uint32_t maxNumSubLeaves = EAX;
+			if (type_ & tAVX && (EBX & (1U << 5))) type_ |= tAVX2;
+			if (EBX & (1U << 3)) type_ |= tBMI1;
+			if (EBX & (1U << 8)) type_ |= tBMI2;
+			if (EBX & (1U << 9)) type_ |= tENHANCED_REP;
+			if (EBX & (1U << 18)) type_ |= tRDSEED;
+			if (EBX & (1U << 19)) type_ |= tADX;
+			if (EBX & (1U << 20)) type_ |= tSMAP;
+			if (EBX & (1U << 23)) type_ |= tCLFLUSHOPT;
+			if (EBX & (1U << 4)) type_ |= tHLE;
+			if (EBX & (1U << 11)) type_ |= tRTM;
+			if (EBX & (1U << 14)) type_ |= tMPX;
+			if (EBX & (1U << 29)) type_ |= tSHA;
+			if (ECX & (1U << 0)) type_ |= tPREFETCHWT1;
+			if (ECX & (1U << 5)) type_ |= tWAITPKG;
+			if (ECX & (1U << 8)) type_ |= tGFNI;
+			if (ECX & (1U << 9)) type_ |= tVAES;
+			if (ECX & (1U << 10)) type_ |= tVPCLMULQDQ;
+			if (ECX & (1U << 25)) type_ |= tCLDEMOTE;
+			if (ECX & (1U << 27)) type_ |= tMOVDIRI;
+			if (ECX & (1U << 28)) type_ |= tMOVDIR64B;
+			if (EDX & (1U << 5)) type_ |= tUINTR;
+			if (EDX & (1U << 14)) type_ |= tSERIALIZE;
+			if (EDX & (1U << 22)) type_ |= tAMX_BF16;
+			if (EDX & (1U << 24)) type_ |= tAMX_TILE;
+			if (EDX & (1U << 25)) type_ |= tAMX_INT8;
+			if (maxNumSubLeaves >= 1) {
+				getCpuidEx(7, 1, data);
+				if (EAX & (1U << 0)) type_ |= tSHA512;
+				if (EAX & (1U << 1)) type_ |= tSM3;
+				if (EAX & (1U << 2)) type_ |= tSM4;
+				if (EAX & (1U << 3)) type_ |= tRAO_INT;
+				if (EAX & (1U << 4)) type_ |= tAVX_VNNI;
+				if (type_ & tAVX512F) {
+					if (EAX & (1U << 5)) type_ |= tAVX512_BF16;
+				}
+				if (EAX & (1U << 7)) type_ |= tCMPCCXADD;
+				if (EAX & (1U << 21)) type_ |= tAMX_FP16;
+				if (EAX & (1U << 23)) type_ |= tAVX_IFMA;
+				if (EDX & (1U << 4)) type_ |= tAVX_VNNI_INT8;
+				if (EDX & (1U << 5)) type_ |= tAVX_NE_CONVERT;
+				if (EDX & (1U << 10)) type_ |= tAVX_VNNI_INT16;
+				if (EDX & (1U << 14)) type_ |= tPREFETCHITI;
+			}
 		}
 		setFamily();
+		setNumCores();
+		setCacheHierarchy();
 	}
 	void putFamily() const
 	{
+#ifndef XBYAK_ONLY_CLASS_CPU
 		printf("family=%d, model=%X, stepping=%d, extFamily=%d, extModel=%X\n",
 			family, model, stepping, extFamily, extModel);
 		printf("display:family=%X, model=%X\n", displayFamily, displayModel);
+#endif
 	}
-	bool has(Type type) const
+	bool has(const Type& type) const
 	{
-		return (type & type_) != 0;
+		return (type & type_) == type;
 	}
 };
 
+#ifndef XBYAK_ONLY_CLASS_CPU
 class Clock {
 public:
-	static inline uint64 getRdtsc()
+	static inline uint64_t getRdtsc()
 	{
-#ifdef _MSC_VER
+#ifdef XBYAK_INTEL_CPU_SPECIFIC
+	#ifdef _MSC_VER
 		return __rdtsc();
-#else
-		unsigned int eax, edx;
+	#else
+		uint32_t eax, edx;
 		__asm__ volatile("rdtsc" : "=a"(eax), "=d"(edx));
-		return ((uint64)edx << 32) | eax;
+		return ((uint64_t)edx << 32) | eax;
+	#endif
+#else
+		// TODO: Need another impl of Clock or rdtsc-equivalent for non-x86 cpu
+		return 0;
 #endif
 	}
 	Clock()
@@ -295,10 +680,10 @@ public:
 		count_++;
 	}
 	int getCount() const { return count_; }
-	uint64 getClock() const { return clock_; }
+	uint64_t getClock() const { return clock_; }
 	void clear() { count_ = 0; clock_ = 0; }
 private:
-	uint64 clock_;
+	uint64_t clock_;
 	int count_;
 };
 
@@ -307,8 +692,8 @@ const int UseRCX = 1 << 6;
 const int UseRDX = 1 << 7;
 
 class Pack {
-	static const size_t maxTblNum = 10;
-	const Xbyak::Reg64 *tbl_[maxTblNum];
+	static const size_t maxTblNum = 15;
+	Xbyak::Reg64 tbl_[maxTblNum];
 	size_t n_;
 public:
 	Pack() : tbl_(), n_(0) {}
@@ -316,56 +701,65 @@ public:
 	Pack(const Pack& rhs)
 		: n_(rhs.n_)
 	{
-		if (n_ > maxTblNum) throw Error(ERR_INTERNAL);
 		for (size_t i = 0; i < n_; i++) tbl_[i] = rhs.tbl_[i];
 	}
+	Pack& operator=(const Pack& rhs)
+	{
+		n_ = rhs.n_;
+		for (size_t i = 0; i < n_; i++) tbl_[i] = rhs.tbl_[i];
+		return *this;
+	}
 	Pack(const Xbyak::Reg64& t0)
-	{ n_ = 1; tbl_[0] = &t0; }
+	{ n_ = 1; tbl_[0] = t0; }
 	Pack(const Xbyak::Reg64& t1, const Xbyak::Reg64& t0)
-	{ n_ = 2; tbl_[0] = &t0; tbl_[1] = &t1; }
+	{ n_ = 2; tbl_[0] = t0; tbl_[1] = t1; }
 	Pack(const Xbyak::Reg64& t2, const Xbyak::Reg64& t1, const Xbyak::Reg64& t0)
-	{ n_ = 3; tbl_[0] = &t0; tbl_[1] = &t1; tbl_[2] = &t2; }
+	{ n_ = 3; tbl_[0] = t0; tbl_[1] = t1; tbl_[2] = t2; }
 	Pack(const Xbyak::Reg64& t3, const Xbyak::Reg64& t2, const Xbyak::Reg64& t1, const Xbyak::Reg64& t0)
-	{ n_ = 4; tbl_[0] = &t0; tbl_[1] = &t1; tbl_[2] = &t2; tbl_[3] = &t3; }
+	{ n_ = 4; tbl_[0] = t0; tbl_[1] = t1; tbl_[2] = t2; tbl_[3] = t3; }
 	Pack(const Xbyak::Reg64& t4, const Xbyak::Reg64& t3, const Xbyak::Reg64& t2, const Xbyak::Reg64& t1, const Xbyak::Reg64& t0)
-	{ n_ = 5; tbl_[0] = &t0; tbl_[1] = &t1; tbl_[2] = &t2; tbl_[3] = &t3; tbl_[4] = &t4; }
+	{ n_ = 5; tbl_[0] = t0; tbl_[1] = t1; tbl_[2] = t2; tbl_[3] = t3; tbl_[4] = t4; }
 	Pack(const Xbyak::Reg64& t5, const Xbyak::Reg64& t4, const Xbyak::Reg64& t3, const Xbyak::Reg64& t2, const Xbyak::Reg64& t1, const Xbyak::Reg64& t0)
-	{ n_ = 6; tbl_[0] = &t0; tbl_[1] = &t1; tbl_[2] = &t2; tbl_[3] = &t3; tbl_[4] = &t4; tbl_[5] = &t5; }
+	{ n_ = 6; tbl_[0] = t0; tbl_[1] = t1; tbl_[2] = t2; tbl_[3] = t3; tbl_[4] = t4; tbl_[5] = t5; }
 	Pack(const Xbyak::Reg64& t6, const Xbyak::Reg64& t5, const Xbyak::Reg64& t4, const Xbyak::Reg64& t3, const Xbyak::Reg64& t2, const Xbyak::Reg64& t1, const Xbyak::Reg64& t0)
-	{ n_ = 7; tbl_[0] = &t0; tbl_[1] = &t1; tbl_[2] = &t2; tbl_[3] = &t3; tbl_[4] = &t4; tbl_[5] = &t5; tbl_[6] = &t6; }
+	{ n_ = 7; tbl_[0] = t0; tbl_[1] = t1; tbl_[2] = t2; tbl_[3] = t3; tbl_[4] = t4; tbl_[5] = t5; tbl_[6] = t6; }
 	Pack(const Xbyak::Reg64& t7, const Xbyak::Reg64& t6, const Xbyak::Reg64& t5, const Xbyak::Reg64& t4, const Xbyak::Reg64& t3, const Xbyak::Reg64& t2, const Xbyak::Reg64& t1, const Xbyak::Reg64& t0)
-	{ n_ = 8; tbl_[0] = &t0; tbl_[1] = &t1; tbl_[2] = &t2; tbl_[3] = &t3; tbl_[4] = &t4; tbl_[5] = &t5; tbl_[6] = &t6; tbl_[7] = &t7; }
+	{ n_ = 8; tbl_[0] = t0; tbl_[1] = t1; tbl_[2] = t2; tbl_[3] = t3; tbl_[4] = t4; tbl_[5] = t5; tbl_[6] = t6; tbl_[7] = t7; }
 	Pack(const Xbyak::Reg64& t8, const Xbyak::Reg64& t7, const Xbyak::Reg64& t6, const Xbyak::Reg64& t5, const Xbyak::Reg64& t4, const Xbyak::Reg64& t3, const Xbyak::Reg64& t2, const Xbyak::Reg64& t1, const Xbyak::Reg64& t0)
-	{ n_ = 9; tbl_[0] = &t0; tbl_[1] = &t1; tbl_[2] = &t2; tbl_[3] = &t3; tbl_[4] = &t4; tbl_[5] = &t5; tbl_[6] = &t6; tbl_[7] = &t7; tbl_[8] = &t8; }
+	{ n_ = 9; tbl_[0] = t0; tbl_[1] = t1; tbl_[2] = t2; tbl_[3] = t3; tbl_[4] = t4; tbl_[5] = t5; tbl_[6] = t6; tbl_[7] = t7; tbl_[8] = t8; }
 	Pack(const Xbyak::Reg64& t9, const Xbyak::Reg64& t8, const Xbyak::Reg64& t7, const Xbyak::Reg64& t6, const Xbyak::Reg64& t5, const Xbyak::Reg64& t4, const Xbyak::Reg64& t3, const Xbyak::Reg64& t2, const Xbyak::Reg64& t1, const Xbyak::Reg64& t0)
-	{ n_ = 10; tbl_[0] = &t0; tbl_[1] = &t1; tbl_[2] = &t2; tbl_[3] = &t3; tbl_[4] = &t4; tbl_[5] = &t5; tbl_[6] = &t6; tbl_[7] = &t7; tbl_[8] = &t8; tbl_[9] = &t9; }
+	{ n_ = 10; tbl_[0] = t0; tbl_[1] = t1; tbl_[2] = t2; tbl_[3] = t3; tbl_[4] = t4; tbl_[5] = t5; tbl_[6] = t6; tbl_[7] = t7; tbl_[8] = t8; tbl_[9] = t9; }
+	Pack(const Xbyak::Reg64& ta, const Xbyak::Reg64& t9, const Xbyak::Reg64& t8, const Xbyak::Reg64& t7, const Xbyak::Reg64& t6, const Xbyak::Reg64& t5, const Xbyak::Reg64& t4, const Xbyak::Reg64& t3, const Xbyak::Reg64& t2, const Xbyak::Reg64& t1, const Xbyak::Reg64& t0)
+	{ n_ = 11; tbl_[0] = t0; tbl_[1] = t1; tbl_[2] = t2; tbl_[3] = t3; tbl_[4] = t4; tbl_[5] = t5; tbl_[6] = t6; tbl_[7] = t7; tbl_[8] = t8; tbl_[9] = t9; tbl_[10] = ta; }
+	Pack(const Xbyak::Reg64& tb, const Xbyak::Reg64& ta, const Xbyak::Reg64& t9, const Xbyak::Reg64& t8, const Xbyak::Reg64& t7, const Xbyak::Reg64& t6, const Xbyak::Reg64& t5, const Xbyak::Reg64& t4, const Xbyak::Reg64& t3, const Xbyak::Reg64& t2, const Xbyak::Reg64& t1, const Xbyak::Reg64& t0)
+	{ n_ = 12; tbl_[0] = t0; tbl_[1] = t1; tbl_[2] = t2; tbl_[3] = t3; tbl_[4] = t4; tbl_[5] = t5; tbl_[6] = t6; tbl_[7] = t7; tbl_[8] = t8; tbl_[9] = t9; tbl_[10] = ta; tbl_[11] = tb; }
 	Pack& append(const Xbyak::Reg64& t)
 	{
-		if (n_ == 10) {
+		if (n_ == maxTblNum) {
 			fprintf(stderr, "ERR Pack::can't append\n");
-			throw Error(ERR_BAD_PARAMETER);
+			XBYAK_THROW_RET(ERR_BAD_PARAMETER, *this)
 		}
-		tbl_[n_++] = &t;
+		tbl_[n_++] = t;
 		return *this;
 	}
 	void init(const Xbyak::Reg64 *tbl, size_t n)
 	{
 		if (n > maxTblNum) {
 			fprintf(stderr, "ERR Pack::init bad n=%d\n", (int)n);
-			throw Error(ERR_BAD_PARAMETER);
+			XBYAK_THROW(ERR_BAD_PARAMETER)
 		}
 		n_ = n;
 		for (size_t i = 0; i < n; i++) {
-			tbl_[i] = &tbl[i];
+			tbl_[i] = tbl[i];
 		}
 	}
 	const Xbyak::Reg64& operator[](size_t n) const
 	{
 		if (n >= n_) {
-			fprintf(stderr, "ERR Pack bad n=%d\n", (int)n);
-			throw Error(ERR_BAD_PARAMETER);
+			fprintf(stderr, "ERR Pack bad n=%d(%d)\n", (int)n, (int)n_);
+			XBYAK_THROW_RET(ERR_BAD_PARAMETER, rax)
 		}
-		return *tbl_[n];
+		return tbl_[n];
 	}
 	size_t size() const { return n_; }
 	/*
@@ -376,7 +770,7 @@ public:
 		if (num == size_t(-1)) num = n_ - pos;
 		if (pos + num > n_) {
 			fprintf(stderr, "ERR Pack::sub bad pos=%d, num=%d\n", (int)pos, (int)num);
-			throw Error(ERR_BAD_PARAMETER);
+			XBYAK_THROW_RET(ERR_BAD_PARAMETER, Pack())
 		}
 		Pack pack;
 		pack.n_ = num;
@@ -388,7 +782,7 @@ public:
 	void put() const
 	{
 		for (size_t i = 0; i < n_; i++) {
-			printf("%s ", tbl_[i]->toString());
+			printf("%s ", tbl_[i].toString());
 		}
 		printf("\n");
 	}
@@ -404,6 +798,7 @@ class StackFrame {
 	static const int rcxPos = 3;
 	static const int rdxPos = 2;
 #endif
+	static const int maxRegNum = 14; // maxRegNum = 16 - rsp - rax
 	Xbyak::CodeGenerator *code_;
 	int pNum_;
 	int tNum_;
@@ -413,7 +808,7 @@ class StackFrame {
 	int P_;
 	bool makeEpilog_;
 	Xbyak::Reg64 pTbl_[4];
-	Xbyak::Reg64 tTbl_[10];
+	Xbyak::Reg64 tTbl_[maxRegNum];
 	Pack p_;
 	Pack t_;
 	StackFrame(const StackFrame&);
@@ -425,7 +820,7 @@ public:
 		make stack frame
 		@param sf [in] this
 		@param pNum [in] num of function parameter(0 <= pNum <= 4)
-		@param tNum [in] num of temporary register(0 <= tNum <= 10, with UseRCX, UseRDX)
+		@param tNum [in] num of temporary register(0 <= tNum, with UseRCX, UseRDX) #{pNum + tNum [+rcx] + [rdx]} <= 14
 		@param stackSizeByte [in] local stack size
 		@param makeEpilog [in] automatically call close() if true
 
@@ -450,29 +845,19 @@ public:
 		, t(t_)
 	{
 		using namespace Xbyak;
-		if (pNum < 0 || pNum > 4) throw Error(ERR_BAD_PNUM);
+		if (pNum < 0 || pNum > 4) XBYAK_THROW(ERR_BAD_PNUM)
 		const int allRegNum = pNum + tNum_ + (useRcx_ ? 1 : 0) + (useRdx_ ? 1 : 0);
-		if (allRegNum < pNum || allRegNum > 14) throw Error(ERR_BAD_TNUM);
+		if (tNum_ < 0 || allRegNum > maxRegNum) XBYAK_THROW(ERR_BAD_TNUM)
 		const Reg64& _rsp = code->rsp;
-		const AddressFrame& _ptr = code->ptr;
-		saveNum_ = (std::max)(0, allRegNum - noSaveNum);
+		saveNum_ = local::max_(0, allRegNum - noSaveNum);
 		const int *tbl = getOrderTbl() + noSaveNum;
-		P_ = saveNum_ + (stackSizeByte + 7) / 8;
-		if (P_ > 0 && (P_ & 1) == 0) P_++; // here (rsp % 16) == 8, then increment P_ for 16 byte alignment
-		P_ *= 8;
-		if (P_ > 0) code->sub(_rsp, P_);
-#ifdef XBYAK64_WIN
-		for (int i = 0; i < (std::min)(saveNum_, 4); i++) {
-			code->mov(_ptr [_rsp + P_ + (i + 1) * 8], Reg64(tbl[i]));
-		}
-		for (int i = 4; i < saveNum_; i++) {
-			code->mov(_ptr [_rsp + P_ - 8 * (saveNum_ - i)], Reg64(tbl[i]));
-		}
-#else
 		for (int i = 0; i < saveNum_; i++) {
-			code->mov(_ptr [_rsp + P_ - 8 * (saveNum_ - i)], Reg64(tbl[i]));
+			code->push(Reg64(tbl[i]));
 		}
-#endif
+		P_ = (stackSizeByte + 7) / 8;
+		if (P_ > 0 && (P_ & 1) == (saveNum_ & 1)) P_++; // (rsp % 16) == 8, then increment P_ for 16 byte alignment
+		P_ *= 8;
+		if (P_ > 0) code->sub(_rsp, P_);
 		int pos = 0;
 		for (int i = 0; i < pNum; i++) {
 			pTbl_[i] = Xbyak::Reg64(getRegIdx(pos));
@@ -493,36 +878,18 @@ public:
 	{
 		using namespace Xbyak;
 		const Reg64& _rsp = code_->rsp;
-		const AddressFrame& _ptr = code_->ptr;
 		const int *tbl = getOrderTbl() + noSaveNum;
-#ifdef XBYAK64_WIN
-		for (int i = 0; i < (std::min)(saveNum_, 4); i++) {
-			code_->mov(Reg64(tbl[i]), _ptr [_rsp + P_ + (i + 1) * 8]);
-		}
-		for (int i = 4; i < saveNum_; i++) {
-			code_->mov(Reg64(tbl[i]), _ptr [_rsp + P_ - 8 * (saveNum_ - i)]);
-		}
-#else
+		if (P_ > 0) code_->add(_rsp, P_);
 		for (int i = 0; i < saveNum_; i++) {
-			code_->mov(Reg64(tbl[i]), _ptr [_rsp + P_ - 8 * (saveNum_ - i)]);
+			code_->pop(Reg64(tbl[saveNum_ - 1 - i]));
 		}
-#endif
-		if (P_ > 0) code_->add(_rsp, P_);
 
 		if (callRet) code_->ret();
 	}
 	~StackFrame()
 	{
 		if (!makeEpilog_) return;
-		try {
-			close();
-		} catch (std::exception& e) {
-			printf("ERR:StackFrame %s\n", e.what());
-			exit(1);
-		} catch (...) {
-			printf("ERR:StackFrame otherwise\n");
-			exit(1);
-		}
+		close();
 	}
 private:
 	const int *getOrderTbl() const
@@ -540,7 +907,7 @@ private:
 	}
 	int getRegIdx(int& pos) const
 	{
-		assert(pos < 14);
+		assert(pos < maxRegNum);
 		using namespace Xbyak;
 		const int *tbl = getOrderTbl();
 		int r = tbl[pos++];
@@ -557,5 +924,137 @@ private:
 };
 #endif
 
+class Profiler {
+	int mode_;
+	const char *suffix_;
+	const void *startAddr_;
+#ifdef XBYAK_USE_PERF
+	FILE *fp_;
+#endif
+public:
+	enum {
+		None = 0,
+		Perf = 1,
+		VTune = 2
+	};
+	Profiler()
+		: mode_(None)
+		, suffix_("")
+		, startAddr_(0)
+#ifdef XBYAK_USE_PERF
+		, fp_(0)
+#endif
+	{
+	}
+	// append suffix to funcName
+	void setNameSuffix(const char *suffix)
+	{
+		suffix_ = suffix;
+	}
+	void setStartAddr(const void *startAddr)
+	{
+		startAddr_ = startAddr;
+	}
+	void init(int mode)
+	{
+		mode_ = None;
+		switch (mode) {
+		default:
+		case None:
+			return;
+		case Perf:
+#ifdef XBYAK_USE_PERF
+			close();
+			{
+				const int pid = getpid();
+				char name[128];
+				snprintf(name, sizeof(name), "/tmp/perf-%d.map", pid);
+				fp_ = fopen(name, "a+");
+				if (fp_ == 0) {
+					fprintf(stderr, "can't open %s\n", name);
+					return;
+				}
+			}
+			mode_ = Perf;
+#endif
+			return;
+		case VTune:
+#ifdef XBYAK_USE_VTUNE
+			dlopen("dummy", RTLD_LAZY); // force to load dlopen to enable jit profiling
+			if (iJIT_IsProfilingActive() != iJIT_SAMPLING_ON) {
+				fprintf(stderr, "VTune profiling is not active\n");
+				return;
+			}
+			mode_ = VTune;
+#endif
+			return;
+		}
+	}
+	~Profiler()
+	{
+		close();
+	}
+	void close()
+	{
+#ifdef XBYAK_USE_PERF
+		if (fp_ == 0) return;
+		fclose(fp_);
+		fp_ = 0;
+#endif
+	}
+	void set(const char *funcName, const void *startAddr, size_t funcSize) const
+	{
+		if (mode_ == None) return;
+#if !defined(XBYAK_USE_PERF) && !defined(XBYAK_USE_VTUNE)
+		(void)funcName;
+		(void)startAddr;
+		(void)funcSize;
+#endif
+#ifdef XBYAK_USE_PERF
+		if (mode_ == Perf) {
+			if (fp_ == 0) return;
+			fprintf(fp_, "%llx %zx %s%s", (long long)startAddr, funcSize, funcName, suffix_);
+			/*
+				perf does not recognize the function name which is less than 3,
+				so append '_' at the end of the name if necessary
+			*/
+			size_t n = strlen(funcName) + strlen(suffix_);
+			for (size_t i = n; i < 3; i++) {
+				fprintf(fp_, "_");
+			}
+			fprintf(fp_, "\n");
+			fflush(fp_);
+		}
+#endif
+#ifdef XBYAK_USE_VTUNE
+		if (mode_ != VTune) return;
+		char className[] = "";
+		char fileName[] = "";
+		iJIT_Method_Load jmethod = {};
+		jmethod.method_id = iJIT_GetNewMethodID();
+		jmethod.class_file_name = className;
+		jmethod.source_file_name = fileName;
+		jmethod.method_load_address = const_cast<void*>(startAddr);
+		jmethod.method_size = funcSize;
+		jmethod.line_number_size = 0;
+		char buf[128];
+		snprintf(buf, sizeof(buf), "%s%s", funcName, suffix_);
+		jmethod.method_name = buf;
+		iJIT_NotifyEvent(iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED, (void*)&jmethod);
+#endif
+	}
+	/*
+		for continuous set
+		funcSize = endAddr - <previous set endAddr>
+	*/
+	void set(const char *funcName, const void *endAddr)
+	{
+		set(funcName, startAddr_, (size_t)endAddr - (size_t)startAddr_);
+		startAddr_ = endAddr;
+	}
+};
+#endif // XBYAK_ONLY_CLASS_CPU
+
 } } // end of util
+
 #endif
diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt
index 228c81f0e..cec43dc6f 100644
--- a/src/core/CMakeLists.txt
+++ b/src/core/CMakeLists.txt
@@ -127,7 +127,7 @@ target_link_libraries(core PRIVATE stb xxhash imgui rapidjson)
 
 if(${CPU_ARCH} STREQUAL "x64")
   target_include_directories(core PUBLIC "${CMAKE_CURRENT_SOURCE_DIR}/../../dep/xbyak/xbyak")
-  target_compile_definitions(core PUBLIC "WITH_RECOMPILER=1" "WITH_MMAP_FASTMEM=1")
+  target_compile_definitions(core PUBLIC "XBYAK_NO_EXCEPTION=1" "WITH_RECOMPILER=1" "WITH_MMAP_FASTMEM=1")
   target_sources(core PRIVATE ${RECOMPILER_SRCS}
     cpu_recompiler_code_generator_x64.cpp
   )
diff --git a/src/core/core.props b/src/core/core.props
index f30bb65e5..d4c832dba 100644
--- a/src/core/core.props
+++ b/src/core/core.props
@@ -13,6 +13,8 @@
       <AdditionalIncludeDirectories Condition="'$(Platform)'!='ARM64'">%(AdditionalIncludeDirectories);$(SolutionDir)dep\rainterface</AdditionalIncludeDirectories>
       
       <AdditionalIncludeDirectories Condition="'$(Platform)'=='x64'">%(AdditionalIncludeDirectories);$(SolutionDir)dep\xbyak\xbyak</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions Condition="'$(Platform)'=='x64'">XBYAK_NO_EXCEPTION=1;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+
       <AdditionalIncludeDirectories Condition="'$(Platform)'=='ARM' Or '$(Platform)'=='ARM64'">%(AdditionalIncludeDirectories);$(SolutionDir)dep\vixl\include</AdditionalIncludeDirectories>
     </ClCompile>
   </ItemDefinitionGroup>
diff --git a/src/core/cpu_recompiler_types.h b/src/core/cpu_recompiler_types.h
index eaeaeae81..b12877129 100644
--- a/src/core/cpu_recompiler_types.h
+++ b/src/core/cpu_recompiler_types.h
@@ -66,7 +66,7 @@ enum class Condition : u8
 
 #if defined(CPU_X64)
 
-using HostReg = Xbyak::Operand::Code;
+using HostReg = unsigned;
 using CodeEmitter = Xbyak::CodeGenerator;
 using LabelType = Xbyak::Label;
 enum : u32