[AVX512] Emit select instruction instead of using x86 specific instrinsics.

[igorb]

Summary

[AVX512] Emit select instruction instead of using x86 specific instrinsics.

This will allow us to remove the x86 instrinics from the backend.

include/clang/Basic/BuiltinsX86.def

 TARGET_BUILTIN(__builtin_ia32_pmaxud512_mask, "V16iV16iV16iV16iUs", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_pmaxuq512_mask, "V8LLiV8LLiV8LLiV8LLiUc", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_pminsd512_mask, "V16iV16iV16iV16iUs", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_pminsq512_mask, "V8LLiV8LLiV8LLiV8LLiUc", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_pminud512_mask, "V16iV16iV16iV16iUs", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_pminuq512_mask, "V8LLiV8LLiV8LLiV8LLiUc", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_pmuldq512_mask, "V8LLiV16iV16iV8LLiUc", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_pmuludq512_mask, "V8LLiV16iV16iV8LLiUc", "", "avx512f")
-TARGET_BUILTIN(__builtin_ia32_blendmd_512_mask, "V16iV16iV16iUs", "", "avx512f")
-TARGET_BUILTIN(__builtin_ia32_blendmq_512_mask, "V8LLiV8LLiV8LLiUc", "", "avx512f")
-TARGET_BUILTIN(__builtin_ia32_blendmps_512_mask, "V16fV16fV16fUs", "", "avx512f")
-TARGET_BUILTIN(__builtin_ia32_blendmpd_512_mask, "V8dV8dV8dUc", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_ptestmd512, "UsV16iV16iUs", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_ptestmq512, "UcV8LLiV8LLiUc", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_pbroadcastd512, "V16iV4iV16iUs","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_pbroadcastq512, "V8LLiV2LLiV8LLiUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_pbroadcastd512_gpr_mask, "V16iiV16iUs", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_pbroadcastq512_gpr_mask, "V8LLiLLiV8LLiUc", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_pbroadcastq512_mem_mask, "V8LLiLLiV8LLiUc", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_loaddqusi512_mask, "V16iiC*V16iUs", "", "avx512f")

 TARGET_BUILTIN(__builtin_ia32_xorpd128_mask, "V2dV2dV2dV2dUc", "", "avx512vl,avx512dq")
 TARGET_BUILTIN(__builtin_ia32_xorps256_mask, "V8fV8fV8fV8fUc", "", "avx512vl,avx512dq")
 TARGET_BUILTIN(__builtin_ia32_xorps128_mask, "V4fV4fV4fV4fUc", "", "avx512vl,avx512dq")
 TARGET_BUILTIN(__builtin_ia32_orpd256_mask, "V4dV4dV4dV4dUc", "", "avx512vl,avx512dq")
 TARGET_BUILTIN(__builtin_ia32_orpd128_mask, "V2dV2dV2dV2dUc", "", "avx512vl,avx512dq")
 TARGET_BUILTIN(__builtin_ia32_orps256_mask, "V8fV8fV8fV8fUc", "", "avx512vl,avx512dq")
 TARGET_BUILTIN(__builtin_ia32_orps128_mask, "V4fV4fV4fV4fUc", "", "avx512vl,avx512dq")
 
-TARGET_BUILTIN(__builtin_ia32_blendmb_512_mask, "V64cV64cV64cULLi", "", "avx512bw")
-TARGET_BUILTIN(__builtin_ia32_blendmw_512_mask, "V32sV32sV32sUi", "", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_pabsb512_mask, "V64cV64cV64cULLi", "", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_pabsw512_mask, "V32sV32sV32sUi", "", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_packssdw512_mask, "V32sV16iV16iV32sUi", "", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_packsswb512_mask, "V64cV32sV32sV64cULLi", "", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_packusdw512_mask, "V32sV16iV16iV32sUi", "", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_packuswb512_mask, "V64cV32sV32sV64cULLi", "", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_paddsb512_mask, "V64cV64cV64cV64cULLi", "", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_paddsw512_mask, "V32sV32sV32sV32sUi", "", "avx512bw")

 TARGET_BUILTIN(__builtin_ia32_vpconflictdi_256_mask, "V4LLiV4LLiV4LLiUc","","avx512cd,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_vpconflictsi_128_mask, "V4iV4iV4iUc","","avx512cd,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_vpconflictsi_256_mask, "V8iV8iV8iUc","","avx512cd,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_vpconflictdi_512_mask, "V8LLiV8LLiV8LLiUc", "", "avx512cd")
 TARGET_BUILTIN(__builtin_ia32_vpconflictsi_512_mask, "V16iV16iV16iUs", "", "avx512cd")
 TARGET_BUILTIN(__builtin_ia32_vplzcntd_512_mask, "V16iV16iV16iUs", "", "avx512cd")
 TARGET_BUILTIN(__builtin_ia32_vplzcntq_512_mask, "V8LLiV8LLiV8LLiUc", "", "avx512cd")
 
-TARGET_BUILTIN(__builtin_ia32_blendmb_128_mask, "V16cV16cV16cUs", "", "avx512vl,avx512bw")
-TARGET_BUILTIN(__builtin_ia32_blendmb_256_mask, "V32cV32cV32cUi", "", "avx512vl,avx512bw")
-TARGET_BUILTIN(__builtin_ia32_blendmw_128_mask, "V8sV8sV8sUc", "", "avx512vl,avx512bw")
-TARGET_BUILTIN(__builtin_ia32_blendmw_256_mask, "V16sV16sV16sUs", "", "avx512vl,avx512bw")
 TARGET_BUILTIN(__builtin_ia32_pabsb128_mask, "V16cV16cV16cUs", "", "avx512vl,avx512bw")
 TARGET_BUILTIN(__builtin_ia32_pabsb256_mask, "V32cV32cV32cUi", "", "avx512vl,avx512bw")
 TARGET_BUILTIN(__builtin_ia32_pabsw128_mask, "V8sV8sV8sUc", "", "avx512vl,avx512bw")
 TARGET_BUILTIN(__builtin_ia32_pabsw256_mask, "V16sV16sV16sUs", "", "avx512vl,avx512bw")
 TARGET_BUILTIN(__builtin_ia32_packssdw128_mask, "V8sV4iV4iV8sUc", "", "avx512vl,avx512bw")
 TARGET_BUILTIN(__builtin_ia32_packssdw256_mask, "V16sV8iV8iV16sUs", "", "avx512vl,avx512bw")
 TARGET_BUILTIN(__builtin_ia32_packsswb128_mask, "V16cV8sV8sV16cUs", "", "avx512vl,avx512bw")
 TARGET_BUILTIN(__builtin_ia32_packsswb256_mask, "V32cV16sV16sV32cUi", "", "avx512vl,avx512bw")

 TARGET_BUILTIN(__builtin_ia32_subsd_round_mask, "V2dV2dV2dV2dUcIi", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_maxsd_round_mask, "V2dV2dV2dV2dUcIi", "", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_minsd_round_mask, "V2dV2dV2dV2dUcIi", "", "avx512f")
 
 TARGET_BUILTIN(__builtin_ia32_addpd128_mask, "V2dV2dV2dV2dUc", "", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_addpd256_mask, "V4dV4dV4dV4dUc", "", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_addps128_mask, "V4fV4fV4fV4fUc", "", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_addps256_mask, "V8fV8fV8fV8fUc", "", "avx512vl")
-TARGET_BUILTIN(__builtin_ia32_blendmd_128_mask, "V4iV4iV4iUc", "", "avx512vl")
-TARGET_BUILTIN(__builtin_ia32_blendmd_256_mask, "V8iV8iV8iUc", "", "avx512vl")
-TARGET_BUILTIN(__builtin_ia32_blendmpd_128_mask, "V2dV2dV2dUc", "", "avx512vl")
-TARGET_BUILTIN(__builtin_ia32_blendmpd_256_mask, "V4dV4dV4dUc", "", "avx512vl")
-TARGET_BUILTIN(__builtin_ia32_blendmps_128_mask, "V4fV4fV4fUc", "", "avx512vl")
-TARGET_BUILTIN(__builtin_ia32_blendmps_256_mask, "V8fV8fV8fUc", "", "avx512vl")
-TARGET_BUILTIN(__builtin_ia32_blendmq_128_mask, "V2LLiV2LLiV2LLiUc", "", "avx512vl")
-TARGET_BUILTIN(__builtin_ia32_blendmq_256_mask, "V4LLiV4LLiV4LLiUc", "", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_compressdf128_mask, "V2dV2dV2dUc", "", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_compressdf256_mask, "V4dV4dV4dUc", "", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_compressdi128_mask, "V2LLiV2LLiV2LLiUc", "", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_compressdi256_mask, "V4LLiV4LLiV4LLiUc", "", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_compresssf128_mask, "V4fV4fV4fUc", "", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_compresssf256_mask, "V8fV8fV8fUc", "", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_compresssi128_mask, "V4iV4iV4iUc", "", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_compresssi256_mask, "V8iV8iV8iUc", "", "avx512vl")

 TARGET_BUILTIN(__builtin_ia32_psrawi128_mask, "V8sV8sIiV8sUc","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_psrawi256_mask, "V16sV16sIiV16sUs","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_psrlw512_mask, "V32sV32sV8sV32sUi","","avx512bw")
 TARGET_BUILTIN(__builtin_ia32_psrlwi512_mask, "V32sV32sIiV32sUi","","avx512bw")
 TARGET_BUILTIN(__builtin_ia32_psrlw128_mask, "V8sV8sV8sV8sUc","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_psrlw256_mask, "V16sV16sV8sV16sUs","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_psrlwi128_mask, "V8sV8sIiV8sUc","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_psrlwi256_mask, "V16sV16sIiV16sUs","","avx512bw,avx512vl")
-TARGET_BUILTIN(__builtin_ia32_movdqa32_128_mask, "V4iV4iV4iUc","","avx512vl")
-TARGET_BUILTIN(__builtin_ia32_movdqa32_256_mask, "V8iV8iV8iUc","","avx512vl")
-TARGET_BUILTIN(__builtin_ia32_movdqa32_512_mask, "V16iV16iV16iUs","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_movdqa32load128_mask, "V4iV4i*V4iUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_movdqa32load256_mask, "V8iV8i*V8iUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_movdqa32load512_mask, "V16iV16iC*V16iUs","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_movdqa32store512_mask, "vV16i*V16iUs","","avx512f")
-TARGET_BUILTIN(__builtin_ia32_movdqa64_512_mask, "V8LLiV8LLiV8LLiUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_movdqa64load512_mask, "V8LLiV8LLiC*V8LLiUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_movdqa64store512_mask, "vV8LLi*V8LLiUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_movdqa32store128_mask, "vV4i*V4iUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_movdqa32store256_mask, "vV8i*V8iUc","","avx512f")
-TARGET_BUILTIN(__builtin_ia32_movdqa64_128_mask, "V2LLiV2LLiV2LLiUc","","avx512vl")
-TARGET_BUILTIN(__builtin_ia32_movdqa64_256_mask, "V4LLiV4LLiV4LLiUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_movdqa64load128_mask, "V2LLiV2LLiC*V2LLiUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_movdqa64load256_mask, "V4LLiV4LLiC*V4LLiUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_movdqa64store128_mask, "vV2LLi*V2LLiUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_movdqa64store256_mask, "vV4LLi*V4LLiUc","","avx512f")
-TARGET_BUILTIN(__builtin_ia32_movdquhi512_mask, "V32sV32sV32sUi","","avx512bw")
-TARGET_BUILTIN(__builtin_ia32_movdquqi512_mask, "V64cV64cV64cULLi","","avx512bw")
-TARGET_BUILTIN(__builtin_ia32_movdquhi128_mask, "V8sV8sV8sUc","","avx512bw,avx512vl")
-TARGET_BUILTIN(__builtin_ia32_movdquhi256_mask, "V16sV16sV16sUs","","avx512bw,avx512vl")
-TARGET_BUILTIN(__builtin_ia32_movdquqi128_mask, "V16cV16cV16cUs","","avx512bw,avx512vl")
-TARGET_BUILTIN(__builtin_ia32_movdquqi256_mask, "V32cV32cV32cUi","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_movddup512_mask, "V8dV8dV8dUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_movddup128_mask, "V2dV2dV2dUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_movddup256_mask, "V4dV4dV4dUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_pbroadcastb512_gpr_mask, "V64ccV64cULLi","","avx512bw")
 TARGET_BUILTIN(__builtin_ia32_pbroadcastb128_gpr_mask, "V16ccV16cUs","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_pbroadcastb256_gpr_mask, "V32ccV32cUi","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_pbroadcastd128_gpr_mask, "V4iiV4iUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_pbroadcastd256_gpr_mask, "V8iiV8iUc","","avx512vl")

 TARGET_BUILTIN(__builtin_ia32_expanddi512_mask, "V8LLiV8LLiV8LLiUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_expandloaddf512_mask, "V8dV8dC*V8dUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_expandloaddi512_mask, "V8LLiV8LLiC*V8LLiUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_expandloadsf512_mask, "V16fV16fC*V16fUs","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_expandloadsi512_mask, "V16iV16iC*V16iUs","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_expandsf512_mask, "V16fV16fV16fUs","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_expandsi512_mask, "V16iV16iV16iUs","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_cvtps2pd512_mask, "V8dV8fV8dUcIi","","avx512f")
-TARGET_BUILTIN(__builtin_ia32_movapd512_mask, "V8dV8dV8dUc","","avx512f")
-TARGET_BUILTIN(__builtin_ia32_movaps512_mask, "V16fV16fV16fUs","","avx512f")
-TARGET_BUILTIN(__builtin_ia32_movapd128_mask, "V2dV2dV2dUc","","avx512vl")
-TARGET_BUILTIN(__builtin_ia32_movapd256_mask, "V4dV4dV4dUc","","avx512vl")
-TARGET_BUILTIN(__builtin_ia32_movaps128_mask, "V4fV4fV4fUc","","avx512vl")
-TARGET_BUILTIN(__builtin_ia32_movaps256_mask, "V8fV8fV8fUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_compressstoredf512_mask, "vV8d*V8dUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_compressstoredi512_mask, "vV8LLi*V8LLiUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_compressstoresf512_mask, "vV16f*V16fUs","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_compressstoresi512_mask, "vV16i*V16iUs","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_vcvtph2ps_mask, "V4fV8sV4fUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_vcvtph2ps256_mask, "V8fV8sV8fUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_vcvtps2ph_mask, "V8sV4fIiV8sUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_vcvtps2ph256_mask, "V8sV8fIiV8sUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_cvtw2mask512, "UiV32s","","avx512bw")
 TARGET_BUILTIN(__builtin_ia32_cvtw2mask128, "UcV8s","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_cvtw2mask256, "UsV16s","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_cvtsd2ss_round_mask, "V4fV4fV2dV4fUcIi","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_cvtsi2sd64, "V2dV2dLLiIi","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_cvtsi2ss32, "V4fV4fiIi","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_cvtsi2ss64, "V4fV4fLLiIi","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_cvtss2sd_round_mask, "V2dV2dV4fV2dUcIi","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_cvtusi2sd32, "V2dV2dUi","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_cvtusi2sd64, "V2dV2dULLiIi","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_cvtusi2ss32, "V4fV4fUiIi","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_cvtusi2ss64, "V4fV4fULLiIi","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_vpmultishiftqb512_mask, "V64cV64cV64cV64cULLi","","avx512vbmi")
 TARGET_BUILTIN(__builtin_ia32_vpmultishiftqb128_mask, "V16cV16cV16cV16cUs","","avx512vbmi,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_vpmultishiftqb256_mask, "V32cV32cV32cV32cUi","","avx512vbmi,avx512vl")
 
+// generic select intrinsics
+TARGET_BUILTIN(__builtin_ia32_selectb_128, "V16cUsV16cV16c", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectb_256, "V32cUiV32cV32c", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectb_512, "V64cULLiV64cV64c", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectw_128, "V8sUcV8sV8s", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectw_256, "V16sUsV16sV16s", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectw_512, "V32sUiV32sV32s", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectd_128, "V4iUcV4iV4i", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectd_256, "V8iUcV8iV8i", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectd_512, "V16iUsV16iV16i", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectq_128, "V2LLiUcV2LLiV2LLi", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectq_256, "V4LLiUcV4LLiV4LLi", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectq_512, "V8LLiUcV8LLiV8LLi", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectps_128, "V4fUcV4fV4f", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectps_256, "V8fUcV8fV8f", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectps_512, "V16fUsV16fV16f", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectpd_128, "V2dUcV2dV2d", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectpd_256, "V4dUcV4dV4d", "", "")
+TARGET_BUILTIN(__builtin_ia32_selectpd_512, "V8dUcV8dV8d", "", "")
+
 // MONITORX/MWAITX
 TARGET_BUILTIN(__builtin_ia32_monitorx, "vv*UiUi", "", "mwaitx")
 TARGET_BUILTIN(__builtin_ia32_mwaitx, "vUiUiUi", "", "mwaitx")
 
 #undef BUILTIN
 #undef TARGET_BUILTIN

lib/CodeGen/CGBuiltin.cpp

Value *Result =
     llvm::UndefValue::get(llvm::VectorType::get(Ops[0]->getType(), Ops.size()));
 
   for (unsigned i = 0, e = Ops.size(); i != e; ++i)
     Result = Builder.CreateInsertElement(Result, Ops[i], Builder.getInt32(i));
 
   return Result;
 }
 
-static Value *EmitX86MaskedStore(CodeGenFunction &CGF,
-                                 SmallVectorImpl<Value *> &Ops,
-                                 unsigned Align) {
-  // Cast the pointer to right type.
-  Ops[0] = CGF.Builder.CreateBitCast(Ops[0],
-                               llvm::PointerType::getUnqual(Ops[1]->getType()));
-
-  // If the mask is all ones just emit a regular store.
-  if (const auto *C = dyn_cast<Constant>(Ops[2]))
-    if (C->isAllOnesValue())
-      return CGF.Builder.CreateAlignedStore(Ops[1], Ops[0], Align);
-
-  // Convert the mask from an integer type to a vector of i1.
-  unsigned NumElts = Ops[1]->getType()->getVectorNumElements();
+static Value *getMaskVecValue(CodeGenFunction &CGF, Value *Mask,
+                              unsigned NumElts) {
+
   llvm::VectorType *MaskTy = llvm::VectorType::get(CGF.Builder.getInt1Ty(),
-                         cast<IntegerType>(Ops[2]->getType())->getBitWidth());
-  Ops[2] = CGF.Builder.CreateBitCast(Ops[2], MaskTy);
+                         cast<IntegerType>(Mask->getType())->getBitWidth());
+  Value *MaskVec = CGF.Builder.CreateBitCast(Mask, MaskTy);
 
   // If we have less than 8 elements, then the starting mask was an i8 and
   // we need to extract down to the right number of elements.
   if (NumElts < 8) {
     int Indices[4];
     for (unsigned i = 0; i != NumElts; ++i)
       Indices[i] = i;
-    Ops[2] = CGF.Builder.CreateShuffleVector(Ops[2], Ops[2],
+    MaskVec = CGF.Builder.CreateShuffleVector(MaskVec, MaskVec,
                                              makeArrayRef(Indices, NumElts),
                                              "extract");
   }
+  return MaskVec;
+}
+
+static Value *EmitX86MaskedStore(CodeGenFunction &CGF,
+                                 SmallVectorImpl<Value *> &Ops,
+                                 unsigned Align) {
+  // Cast the pointer to right type.
+  Ops[0] = CGF.Builder.CreateBitCast(Ops[0],
+                               llvm::PointerType::getUnqual(Ops[1]->getType()));
 
-  return CGF.Builder.CreateMaskedStore(Ops[1], Ops[0], Align, Ops[2]);
+  // If the mask is all ones just emit a regular store.
+  if (const auto *C = dyn_cast<Constant>(Ops[2]))
+    if (C->isAllOnesValue())
+      return CGF.Builder.CreateAlignedStore(Ops[1], Ops[0], Align);
+
+  Value *MaskVec = getMaskVecValue(CGF, Ops[2],
+                                   Ops[1]->getType()->getVectorNumElements());
+
+  return CGF.Builder.CreateMaskedStore(Ops[1], Ops[0], Align, MaskVec);
 }
 
 static Value *EmitX86MaskedLoad(CodeGenFunction &CGF,
                                 SmallVectorImpl<Value *> &Ops, unsigned Align) {
   // Cast the pointer to right type.
   Ops[0] = CGF.Builder.CreateBitCast(Ops[0],
                                llvm::PointerType::getUnqual(Ops[1]->getType()));
 
   // If the mask is all ones just emit a regular store.
   if (const auto *C = dyn_cast<Constant>(Ops[2]))
     if (C->isAllOnesValue())
       return CGF.Builder.CreateAlignedLoad(Ops[0], Align);
 
-  // Convert the mask from an integer type to a vector of i1.
-  unsigned NumElts = Ops[1]->getType()->getVectorNumElements();
-  llvm::VectorType *MaskTy = llvm::VectorType::get(CGF.Builder.getInt1Ty(),
-                         cast<IntegerType>(Ops[2]->getType())->getBitWidth());
-  Ops[2] = CGF.Builder.CreateBitCast(Ops[2], MaskTy);
+  Value *MaskVec = getMaskVecValue(CGF, Ops[2],
+                                   Ops[1]->getType()->getVectorNumElements());
 
-  // If we have less than 8 elements, then the starting mask was an i8 and
-  // we need to extract down to the right number of elements.
-  if (NumElts < 8) {
-    int Indices[4];
-    for (unsigned i = 0; i != NumElts; ++i)
-      Indices[i] = i;
-    Ops[2] = CGF.Builder.CreateShuffleVector(Ops[2], Ops[2],
-                                             makeArrayRef(Indices, NumElts),
-                                             "extract");
-  }
+  return CGF.Builder.CreateMaskedLoad(Ops[0], Align, MaskVec, Ops[1]);
 
-  return CGF.Builder.CreateMaskedLoad(Ops[0], Align, Ops[2], Ops[1]);
+static Value *EmitX86Select(CodeGenFunction &CGF,
+                            SmallVectorImpl<Value *> &Ops) {
+
+  // If the mask is all ones just return first argument.
+  if (const auto *C = dyn_cast<Constant>(Ops[0]))
+    if (C->isAllOnesValue())
+      return Ops[1];
+
+  Value *MaskVec = getMaskVecValue(CGF, Ops[0],
+                                   Ops[1]->getType()->getVectorNumElements());
+
+  return CGF.Builder.CreateSelect(MaskVec, Ops[1], Ops[2]);
 }
 
 Value *CodeGenFunction::EmitX86BuiltinExpr(unsigned BuiltinID,
                                            const CallExpr *E) {
   if (BuiltinID == X86::BI__builtin_ms_va_start ||
       BuiltinID == X86::BI__builtin_ms_va_end)
     return EmitVAStartEnd(EmitMSVAListRef(E->getArg(0)).getPointer(),
                           BuiltinID == X86::BI__builtin_ms_va_start);
case X86::BI__builtin_ia32_movnti64: {
     unsigned Align;
     if (ArgTy->isIntegerType())
       Align = 1;
     else
       Align = getContext().getTypeSizeInChars(ArgTy).getQuantity();
     SI->setAlignment(Align);
     return SI;
   }
+  case X86::BI__builtin_ia32_selectb_128:
+  case X86::BI__builtin_ia32_selectb_256:
+  case X86::BI__builtin_ia32_selectb_512:
+  case X86::BI__builtin_ia32_selectw_128:
+  case X86::BI__builtin_ia32_selectw_256:
+  case X86::BI__builtin_ia32_selectw_512:
+  case X86::BI__builtin_ia32_selectd_128:
+  case X86::BI__builtin_ia32_selectd_256:
+  case X86::BI__builtin_ia32_selectd_512:
+  case X86::BI__builtin_ia32_selectq_128:
+  case X86::BI__builtin_ia32_selectq_256:
+  case X86::BI__builtin_ia32_selectq_512:
+  case X86::BI__builtin_ia32_selectps_128:
+  case X86::BI__builtin_ia32_selectps_256:
+  case X86::BI__builtin_ia32_selectps_512:
+  case X86::BI__builtin_ia32_selectpd_128:
+  case X86::BI__builtin_ia32_selectpd_256:
+  case X86::BI__builtin_ia32_selectpd_512:
+    return EmitX86Select(*this, Ops);
   // 3DNow!
   case X86::BI__builtin_ia32_pswapdsf:
   case X86::BI__builtin_ia32_pswapdsi: {
     llvm::Type *MMXTy = llvm::Type::getX86_MMXTy(getLLVMContext());
     Ops[0] = Builder.CreateBitCast(Ops[0], MMXTy, "cast");
     llvm::Function *F = CGM.getIntrinsic(Intrinsic::x86_3dnowa_pswapd);
     return Builder.CreateCall(F, Ops, "pswapd");
   }

lib/Headers/avx512bwintrin.h

return (__m512i) __builtin_ia32_pmullw512_mask ((__v32hi) __A,
               (__v32hi) __B,
               (__v32hi) _mm512_setzero_hi(),
               (__mmask32) __U);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_blend_epi8 (__mmask64 __U, __m512i __A, __m512i __W)
 {
-  return (__m512i) __builtin_ia32_blendmb_512_mask ((__v64qi) __A,
+  return (__m512i) __builtin_ia32_selectb_512 ((__mmask64) __U,
               (__v64qi) __W,
-              (__mmask64) __U);
+              (__v64qi) __A);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_blend_epi16 (__mmask32 __U, __m512i __A, __m512i __W)
 {
-  return (__m512i) __builtin_ia32_blendmw_512_mask ((__v32hi) __A,
+  return (__m512i) __builtin_ia32_selectw_512 ((__mmask32) __U,
               (__v32hi) __W,
-              (__mmask32) __U);
+              (__v32hi) __A);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_abs_epi8 (__m512i __A)
 {
   return (__m512i) __builtin_ia32_pabsb512_mask ((__v64qi) __A,
               (__v64qi) _mm512_setzero_qi(),
               (__mmask64) -1);
(__m512i)__builtin_ia32_psrlwi512_mask((__v32hi)(__m512i)(A), (int)(imm), \
                                          (__v32hi)_mm512_setzero_hi(), \
                                          (__mmask32)(U)); })
 
 
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_mov_epi16 (__m512i __W, __mmask32 __U, __m512i __A)
 {
-  return (__m512i) __builtin_ia32_movdquhi512_mask ((__v32hi) __A,
-                (__v32hi) __W,
-                (__mmask32) __U);
+  return (__m512i) __builtin_ia32_selectw_512 ((__mmask32) __U,
+                (__v32hi) __A,
+                (__v32hi) __W);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_maskz_mov_epi16 (__mmask32 __U, __m512i __A)
 {
-  return (__m512i) __builtin_ia32_movdquhi512_mask ((__v32hi) __A,
-                (__v32hi)
-                _mm512_setzero_hi (),
-                (__mmask32) __U);
+  return (__m512i) __builtin_ia32_selectw_512 ((__mmask32) __U,
+                (__v32hi) __A,
+                (__v32hi) _mm512_setzero_hi ());
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_mov_epi8 (__m512i __W, __mmask64 __U, __m512i __A)
 {
-  return (__m512i) __builtin_ia32_movdquqi512_mask ((__v64qi) __A,
-                (__v64qi) __W,
-                (__mmask64) __U);
+  return (__m512i) __builtin_ia32_selectb_512 ((__mmask64) __U,
+                (__v64qi) __A,
+                (__v64qi) __W);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_maskz_mov_epi8 (__mmask64 __U, __m512i __A)
 {
-  return (__m512i) __builtin_ia32_movdquqi512_mask ((__v64qi) __A,
-                (__v64qi)
-                _mm512_setzero_hi (),
-                (__mmask64) __U);
+  return (__m512i) __builtin_ia32_selectb_512 ((__mmask64) __U,
+                (__v64qi) __A,
+                (__v64qi) _mm512_setzero_hi ());
 }
 
-
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_set1_epi8 (__m512i __O, __mmask64 __M, char __A)
 {
   return (__m512i) __builtin_ia32_pbroadcastb512_gpr_mask (__A,
                  (__v64qi) __O,
                  __M);
 }
 

lib/Headers/avx512fintrin.h

#define _mm512_maskz_extractf32x4_ps(U, A, imm) __extension__ ({\
   (__m128)__builtin_ia32_extractf32x4_mask((__v16sf)(__m512)(A), (int)(imm), \
                                            (__v4sf)_mm_setzero_ps(), \
                                            (__mmask8)(U)); })
 /* Vector Blend */
 
 static __inline __m512d __DEFAULT_FN_ATTRS
 _mm512_mask_blend_pd(__mmask8 __U, __m512d __A, __m512d __W)
 {
-  return (__m512d) __builtin_ia32_blendmpd_512_mask ((__v8df) __A,
+  return (__m512d) __builtin_ia32_selectpd_512 ((__mmask8) __U,
                  (__v8df) __W,
-                 (__mmask8) __U);
+                 (__v8df) __A);
 }
 
 static __inline __m512 __DEFAULT_FN_ATTRS
 _mm512_mask_blend_ps(__mmask16 __U, __m512 __A, __m512 __W)
 {
-  return (__m512) __builtin_ia32_blendmps_512_mask ((__v16sf) __A,
+  return (__m512) __builtin_ia32_selectps_512 ((__mmask16) __U,
                 (__v16sf) __W,
-                (__mmask16) __U);
+                (__v16sf) __A);
 }
 
 static __inline __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_blend_epi64(__mmask8 __U, __m512i __A, __m512i __W)
 {
-  return (__m512i) __builtin_ia32_blendmq_512_mask ((__v8di) __A,
+  return (__m512i) __builtin_ia32_selectq_512 ((__mmask8) __U,
                 (__v8di) __W,
-                (__mmask8) __U);
+                (__v8di) __A);
 }
 
 static __inline __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_blend_epi32(__mmask16 __U, __m512i __A, __m512i __W)
 {
-  return (__m512i) __builtin_ia32_blendmd_512_mask ((__v16si) __A,
+  return (__m512i) __builtin_ia32_selectd_512 ((__mmask16) __U,
                 (__v16si) __W,
-                (__mmask16) __U);
+                (__v16si) __A);
 }
 
 /* Compare */
 
 #define _mm512_cmp_round_ps_mask(A, B, P, R) __extension__ ({ \
   (__mmask16)__builtin_ia32_cmpps512_mask((__v16sf)(__m512)(A), \
                                           (__v16sf)(__m512)(B), (int)(P), \
                                           (__mmask16)-1, (int)(R)); })

 {
   __builtin_ia32_movdqa32store512_mask ((__v16si *) __P, (__v16si) __A,
           (__mmask16) __U);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_mov_epi32 (__m512i __W, __mmask16 __U, __m512i __A)
 {
-  return (__m512i) __builtin_ia32_movdqa32_512_mask ((__v16si) __A,
-                 (__v16si) __W,
-                 (__mmask16) __U);
+  return (__m512i) __builtin_ia32_selectd_512 ((__mmask16) __U,
+                 (__v16si) __A,
+                 (__v16si) __W);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_maskz_mov_epi32 (__mmask16 __U, __m512i __A)
 {
-  return (__m512i) __builtin_ia32_movdqa32_512_mask ((__v16si) __A,
-                 (__v16si)
-                 _mm512_setzero_si512 (),
-                 (__mmask16) __U);
+  return (__m512i) __builtin_ia32_selectd_512 ((__mmask16) __U,
+                 (__v16si) __A,
+                 (__v16si) _mm512_setzero_si512 ());
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_mov_epi64 (__m512i __W, __mmask8 __U, __m512i __A)
 {
-  return (__m512i) __builtin_ia32_movdqa64_512_mask ((__v8di) __A,
-                 (__v8di) __W,
-                 (__mmask8) __U);
+  return (__m512i) __builtin_ia32_selectq_512 ((__mmask8) __U,
+                 (__v8di) __A,
+                 (__v8di) __W);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_maskz_mov_epi64 (__mmask8 __U, __m512i __A)
 {
-  return (__m512i) __builtin_ia32_movdqa64_512_mask ((__v8di) __A,
-                 (__v8di)
-                 _mm512_setzero_si512 (),
-                 (__mmask8) __U);
+  return (__m512i) __builtin_ia32_selectq_512 ((__mmask8) __U,
+                 (__v8di) __A,
+                 (__v8di) _mm512_setzero_si512 ());
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_load_epi64 (__m512i __W, __mmask8 __U, void const *__P)
 {
   return (__m512i) __builtin_ia32_movdqa64load512_mask ((const __v8di *) __P,
               (__v8di) __W,
               (__mmask8) __U);
return (__m512d) __builtin_ia32_cvtps2pd512_mask ((__v8sf) __A,
                 _mm512_setzero_pd (),
                 (__mmask8) __U,
                 _MM_FROUND_CUR_DIRECTION);
 }
 
 static __inline__ __m512d __DEFAULT_FN_ATTRS
 _mm512_mask_mov_pd (__m512d __W, __mmask8 __U, __m512d __A)
 {
-  return (__m512d) __builtin_ia32_movapd512_mask ((__v8df) __A,
-              (__v8df) __W,
-              (__mmask8) __U);
+  return (__m512d) __builtin_ia32_selectpd_512 ((__mmask8) __U,
+              (__v8df) __A,
+              (__v8df) __W);
 }
 
 static __inline__ __m512d __DEFAULT_FN_ATTRS
 _mm512_maskz_mov_pd (__mmask8 __U, __m512d __A)
 {
-  return (__m512d) __builtin_ia32_movapd512_mask ((__v8df) __A,
-              (__v8df)
-              _mm512_setzero_pd (),
-              (__mmask8) __U);
+  return (__m512d) __builtin_ia32_selectpd_512 ((__mmask8) __U,
+              (__v8df) __A,
+              (__v8df) _mm512_setzero_pd ());
 }
 
 static __inline__ __m512 __DEFAULT_FN_ATTRS
 _mm512_mask_mov_ps (__m512 __W, __mmask16 __U, __m512 __A)
 {
-  return (__m512) __builtin_ia32_movaps512_mask ((__v16sf) __A,
-             (__v16sf) __W,
-             (__mmask16) __U);
+  return (__m512) __builtin_ia32_selectps_512 ((__mmask16) __U,
+             (__v16sf) __A,
+             (__v16sf) __W);
 }
 
 static __inline__ __m512 __DEFAULT_FN_ATTRS
 _mm512_maskz_mov_ps (__mmask16 __U, __m512 __A)
 {
-  return (__m512) __builtin_ia32_movaps512_mask ((__v16sf) __A,
-             (__v16sf)
-             _mm512_setzero_ps (),
-             (__mmask16) __U);
+  return (__m512) __builtin_ia32_selectps_512 ((__mmask16) __U,
+             (__v16sf) __A,
+             (__v16sf) _mm512_setzero_ps ());
 }
 
 static __inline__ void __DEFAULT_FN_ATTRS
 _mm512_mask_compressstoreu_pd (void *__P, __mmask8 __U, __m512d __A)
 {
   __builtin_ia32_compressstoredf512_mask ((__v8df *) __P, (__v8df) __A,
             (__mmask8) __U);
 }

lib/Headers/avx512vlbwintrin.h

return (__m128i) __builtin_ia32_pmullw128_mask ((__v8hi) __A,
               (__v8hi)
               _mm_setzero_si128 (),
               (__mmask8) __U);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_blend_epi8 (__mmask16 __U, __m128i __A, __m128i __W)
 {
-  return (__m128i) __builtin_ia32_blendmb_128_mask ((__v16qi) __A,
-               (__v16qi) __W,
-               (__mmask16) __U);
+  return (__m128i) __builtin_ia32_selectb_128 ((__mmask16) __U,
+              (__v16qi) __A);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mask_blend_epi8 (__mmask32 __U, __m256i __A, __m256i __W)
 {
-  return (__m256i) __builtin_ia32_blendmb_256_mask ((__v32qi) __A,
+  return (__m256i) __builtin_ia32_selectb_256 ((__mmask32) __U,
                (__v32qi) __W,
-               (__mmask32) __U);
+               (__v32qi) __A);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_blend_epi16 (__mmask8 __U, __m128i __A, __m128i __W)
 {
-  return (__m128i) __builtin_ia32_blendmw_128_mask ((__v8hi) __A,
+  return (__m128i) __builtin_ia32_selectw_128 ((__mmask8) __U,
                (__v8hi) __W,
-               (__mmask8) __U);
+               (__v8hi) __A);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mask_blend_epi16 (__mmask16 __U, __m256i __A, __m256i __W)
 {
-  return (__m256i) __builtin_ia32_blendmw_256_mask ((__v16hi) __A,
+  return (__m256i) __builtin_ia32_selectw_256 ((__mmask16) __U,
                (__v16hi) __W,
-               (__mmask16) __U);
+               (__v16hi) __A);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_abs_epi8 (__m128i __W, __mmask16 __U, __m128i __A)
 {
   return (__m128i) __builtin_ia32_pabsb128_mask ((__v16qi) __A,
                (__v16qi) __W,
                (__mmask16) __U);

 #define _mm256_maskz_srli_epi16(U, A, imm) __extension__ ({ \
   (__m256i)__builtin_ia32_psrlwi256_mask((__v16hi)(__m256i)(A), (int)(imm), \
                                          (__v16hi)_mm256_setzero_si256(), \
                                          (__mmask16)(U)); })
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_mov_epi16 (__m128i __W, __mmask8 __U, __m128i __A)
 {
-  return (__m128i) __builtin_ia32_movdquhi128_mask ((__v8hi) __A,
-                (__v8hi) __W,
-                (__mmask8) __U);
+  return (__m128i) __builtin_ia32_selectw_128 ((__mmask8) __U,
+                (__v8hi) __A,
+                (__v8hi) __W);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_maskz_mov_epi16 (__mmask8 __U, __m128i __A)
 {
-  return (__m128i) __builtin_ia32_movdquhi128_mask ((__v8hi) __A,
-                (__v8hi)
-                _mm_setzero_hi (),
-                (__mmask8) __U);
+  return (__m128i) __builtin_ia32_selectw_128 ((__mmask8) __U,
+                (__v8hi) __A,
+                (__v8hi) _mm_setzero_hi ());
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mask_mov_epi16 (__m256i __W, __mmask16 __U, __m256i __A)
 {
-  return (__m256i) __builtin_ia32_movdquhi256_mask ((__v16hi) __A,
-                (__v16hi) __W,
-                (__mmask16) __U);
+  return (__m256i) __builtin_ia32_selectw_256 ((__mmask16) __U,
+                (__v16hi) __A,
+                (__v16hi) __W);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_maskz_mov_epi16 (__mmask16 __U, __m256i __A)
 {
-  return (__m256i) __builtin_ia32_movdquhi256_mask ((__v16hi) __A,
-                (__v16hi)
-                _mm256_setzero_si256 (),
-                (__mmask16) __U);
+  return (__m256i) __builtin_ia32_selectw_256 ((__mmask16) __U,
+                (__v16hi) __A,
+                (__v16hi) _mm256_setzero_si256 ());
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_mov_epi8 (__m128i __W, __mmask16 __U, __m128i __A)
 {
-  return (__m128i) __builtin_ia32_movdquqi128_mask ((__v16qi) __A,
-                (__v16qi) __W,
-                (__mmask16) __U);
+  return (__m128i) __builtin_ia32_selectb_128 ((__mmask16) __U,
+                (__v16qi) __A,
+                (__v16qi) __W);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_maskz_mov_epi8 (__mmask16 __U, __m128i __A)
 {
-  return (__m128i) __builtin_ia32_movdquqi128_mask ((__v16qi) __A,
-                (__v16qi)
-                _mm_setzero_hi (),
-                (__mmask16) __U);
+  return (__m128i) __builtin_ia32_selectb_128 ((__mmask16) __U,
+                (__v16qi) __A,
+                (__v16qi) _mm_setzero_hi ());
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mask_mov_epi8 (__m256i __W, __mmask32 __U, __m256i __A)
 {
-  return (__m256i) __builtin_ia32_movdquqi256_mask ((__v32qi) __A,
-                (__v32qi) __W,
-                (__mmask32) __U);
+  return (__m256i) __builtin_ia32_selectb_256 ((__mmask32) __U,
+                (__v32qi) __A,
+                (__v32qi) __W);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_maskz_mov_epi8 (__mmask32 __U, __m256i __A)
 {
-  return (__m256i) __builtin_ia32_movdquqi256_mask ((__v32qi) __A,
-                (__v32qi)
-                _mm256_setzero_si256 (),
-                (__mmask32) __U);
+  return (__m256i) __builtin_ia32_selectb_256 ((__mmask32) __U,
+                (__v32qi) __A,
+                (__v32qi) _mm256_setzero_si256 ());
 }
 
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_set1_epi8 (__m128i __O, __mmask16 __M, char __A)
 {
   return (__m128i) __builtin_ia32_pbroadcastb128_gpr_mask (__A,
                  (__v16qi) __O,

lib/Headers/avx512vlintrin.h

return (__m256) __builtin_ia32_addps256_mask ((__v8sf) __A,
             (__v8sf) __B,
             (__v8sf)
             _mm256_setzero_ps (),
             (__mmask8) __U);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_blend_epi32 (__mmask8 __U, __m128i __A, __m128i __W) {
-  return (__m128i) __builtin_ia32_blendmd_128_mask ((__v4si) __A,
+  return (__m128i) __builtin_ia32_selectd_128 ((__mmask8) __U,
                 (__v4si) __W,
-                (__mmask8) __U);
+                (__v4si) __A);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mask_blend_epi32 (__mmask8 __U, __m256i __A, __m256i __W) {
-  return (__m256i) __builtin_ia32_blendmd_256_mask ((__v8si) __A,
+  return (__m256i) __builtin_ia32_selectd_256 ((__mmask8) __U,
                 (__v8si) __W,
-                (__mmask8) __U);
+                (__v8si) __A);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_mask_blend_pd (__mmask8 __U, __m128d __A, __m128d __W) {
-  return (__m128d) __builtin_ia32_blendmpd_128_mask ((__v2df) __A,
+  return (__m128d) __builtin_ia32_selectpd_128 ((__mmask8) __U,
                  (__v2df) __W,
-                 (__mmask8) __U);
+                 (__v2df) __A);
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_mask_blend_pd (__mmask8 __U, __m256d __A, __m256d __W) {
-  return (__m256d) __builtin_ia32_blendmpd_256_mask ((__v4df) __A,
+  return (__m256d) __builtin_ia32_selectpd_256 ((__mmask8) __U,
                  (__v4df) __W,
-                 (__mmask8) __U);
+                 (__v4df) __A);
 }
 
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_mask_blend_ps (__mmask8 __U, __m128 __A, __m128 __W) {
-  return (__m128) __builtin_ia32_blendmps_128_mask ((__v4sf) __A,
+  return (__m128) __builtin_ia32_selectps_128 ((__mmask8) __U,
                 (__v4sf) __W,
-                (__mmask8) __U);
+                (__v4sf) __A);
 }
 
 static __inline__ __m256 __DEFAULT_FN_ATTRS
 _mm256_mask_blend_ps (__mmask8 __U, __m256 __A, __m256 __W) {
-  return (__m256) __builtin_ia32_blendmps_256_mask ((__v8sf) __A,
+  return (__m256) __builtin_ia32_selectps_256 ((__mmask8) __U,
                 (__v8sf) __W,
-                (__mmask8) __U);
+                (__v8sf) __A);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_blend_epi64 (__mmask8 __U, __m128i __A, __m128i __W) {
-  return (__m128i) __builtin_ia32_blendmq_128_mask ((__v2di) __A,
+  return (__m128i) __builtin_ia32_selectq_128 ((__mmask8) __U,
                 (__v2di) __W,
-                (__mmask8) __U);
+                (__v2di) __A);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mask_blend_epi64 (__mmask8 __U, __m256i __A, __m256i __W) {
-  return (__m256i) __builtin_ia32_blendmq_256_mask ((__v4di) __A,
+  return (__m256i) __builtin_ia32_selectq_256 ((__mmask8) __U,
                 (__v4di) __W,
-                (__mmask8) __U);
+                (__v4di) __A);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_mask_compress_pd (__m128d __W, __mmask8 __U, __m128d __A) {
   return (__m128d) __builtin_ia32_compressdf128_mask ((__v2df) __A,
                   (__v2df) __W,
                   (__mmask8) __U);
 }
return (__m256i) __builtin_ia32_psravq256_mask ((__v4di) __X,
               (__v4di)
               _mm256_setzero_si256 (),
               (__mmask8) __U);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_mov_epi32 (__m128i __W, __mmask8 __U, __m128i __A)
 {
-  return (__m128i) __builtin_ia32_movdqa32_128_mask ((__v4si) __A,
-                 (__v4si) __W,
-                 (__mmask8) __U);
+  return (__m128i) __builtin_ia32_selectd_128 ((__mmask8) __U,
+                 (__v4si) __A,
+                 (__v4si) __W);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_maskz_mov_epi32 (__mmask8 __U, __m128i __A)
 {
-  return (__m128i) __builtin_ia32_movdqa32_128_mask ((__v4si) __A,
-                 (__v4si)
-                 _mm_setzero_si128 (),
-                 (__mmask8) __U);
+  return (__m128i) __builtin_ia32_selectd_128 ((__mmask8) __U,
+                 (__v4si) __A,
+                 (__v4si) _mm_setzero_si128 ());
 }
 
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mask_mov_epi32 (__m256i __W, __mmask8 __U, __m256i __A)
 {
-  return (__m256i) __builtin_ia32_movdqa32_256_mask ((__v8si) __A,
-                 (__v8si) __W,
-                 (__mmask8) __U);
+  return (__m256i) __builtin_ia32_selectd_256 ((__mmask8) __U,
+                 (__v8si) __A,
+                 (__v8si) __W);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_maskz_mov_epi32 (__mmask8 __U, __m256i __A)
 {
-  return (__m256i) __builtin_ia32_movdqa32_256_mask ((__v8si) __A,
-                 (__v8si)
-                 _mm256_setzero_si256 (),
-                 (__mmask8) __U);
+  return (__m256i) __builtin_ia32_selectd_256 ((__mmask8) __U,
+                 (__v8si) __A,
+                 (__v8si) _mm256_setzero_si256 ());
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_load_epi32 (__m128i __W, __mmask8 __U, void const *__P)
 {
   return (__m128i) __builtin_ia32_movdqa32load128_mask ((__v4si *) __P,
               (__v4si) __W,
               (__mmask8)
_mm256_mask_store_epi32 (void *__P, __mmask8 __U, __m256i __A)
   __builtin_ia32_movdqa32store256_mask ((__v8si *) __P,
           (__v8si) __A,
           (__mmask8) __U);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_mov_epi64 (__m128i __W, __mmask8 __U, __m128i __A)
 {
-  return (__m128i) __builtin_ia32_movdqa64_128_mask ((__v2di) __A,
-                 (__v2di) __W,
-                 (__mmask8) __U);
+  return (__m128i) __builtin_ia32_selectq_128 ((__mmask8) __U,
+                 (__v2di) __A,
+                 (__v2di) __W);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_maskz_mov_epi64 (__mmask8 __U, __m128i __A)
 {
-  return (__m128i) __builtin_ia32_movdqa64_128_mask ((__v2di) __A,
-                 (__v2di)
-                 _mm_setzero_di (),
-                 (__mmask8) __U);
+  return (__m128i) __builtin_ia32_selectq_128 ((__mmask8) __U,
+                 (__v2di) __A,
+                 (__v2di) _mm_setzero_di ());
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mask_mov_epi64 (__m256i __W, __mmask8 __U, __m256i __A)
 {
-  return (__m256i) __builtin_ia32_movdqa64_256_mask ((__v4di) __A,
-                 (__v4di) __W,
-                 (__mmask8) __U);
+  return (__m256i) __builtin_ia32_selectq_256 ((__mmask8) __U,
+                 (__v4di) __A,
+                 (__v4di) __W);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_maskz_mov_epi64 (__mmask8 __U, __m256i __A)
 {
-  return (__m256i) __builtin_ia32_movdqa64_256_mask ((__v4di) __A,
-                 (__v4di)
-                 _mm256_setzero_si256 (),
-                 (__mmask8) __U);
+  return (__m256i) __builtin_ia32_selectq_256 ((__mmask8) __U,
+                 (__v4di) __A,
+                 (__v4di) _mm256_setzero_si256 ());
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_load_epi64 (__m128i __W, __mmask8 __U, void const *__P)
 {
   return (__m128i) __builtin_ia32_movdqa64load128_mask ((__v2di *) __P,
               (__v2di) __W,
               (__mmask8)
#define _mm_maskz_shuffle_epi32(U, A, I) __extension__({\
   (__m128i)__builtin_ia32_pshufd128_mask((__v4si)(__m128i)(A), (int)(I), \
                                          (__v4si)_mm_setzero_si128(), \
                                          (__mmask8)(U)); })
 
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_mask_mov_pd (__m128d __W, __mmask8 __U, __m128d __A)
 {
-  return (__m128d) __builtin_ia32_movapd128_mask ((__v2df) __A,
-              (__v2df) __W,
-              (__mmask8) __U);
+  return (__m128d) __builtin_ia32_selectpd_128 ((__mmask8) __U,
+              (__v2df) __A,
+              (__v2df) __W);
 }
 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_maskz_mov_pd (__mmask8 __U, __m128d __A)
 {
-  return (__m128d) __builtin_ia32_movapd128_mask ((__v2df) __A,
-              (__v2df)
-              _mm_setzero_pd (),
-              (__mmask8) __U);
+  return (__m128d) __builtin_ia32_selectpd_128 ((__mmask8) __U,
+              (__v2df) __A,
+              (__v2df) _mm_setzero_pd ());
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_mask_mov_pd (__m256d __W, __mmask8 __U, __m256d __A)
 {
-  return (__m256d) __builtin_ia32_movapd256_mask ((__v4df) __A,
-              (__v4df) __W,
-              (__mmask8) __U);
+  return (__m256d) __builtin_ia32_selectpd_256 ((__mmask8) __U,
+              (__v4df) __A,
+              (__v4df) __W);
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_maskz_mov_pd (__mmask8 __U, __m256d __A)
 {
-  return (__m256d) __builtin_ia32_movapd256_mask ((__v4df) __A,
-              (__v4df)
-              _mm256_setzero_pd (),
-              (__mmask8) __U);
+  return (__m256d) __builtin_ia32_selectpd_256 ((__mmask8) __U,
+              (__v4df) __A,
+              (__v4df) _mm256_setzero_pd ());
 }
 
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_mask_mov_ps (__m128 __W, __mmask8 __U, __m128 __A)
 {
-  return (__m128) __builtin_ia32_movaps128_mask ((__v4sf) __A,
-             (__v4sf) __W,
-             (__mmask8) __U);
+  return (__m128) __builtin_ia32_selectps_128 ((__mmask8) __U,
+             (__v4sf) __A,
+             (__v4sf) __W);
 }
 
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_maskz_mov_ps (__mmask8 __U, __m128 __A)
 {
-  return (__m128) __builtin_ia32_movaps128_mask ((__v4sf) __A,
-             (__v4sf)
-             _mm_setzero_ps (),
-             (__mmask8) __U);
+  return (__m128) __builtin_ia32_selectps_128 ((__mmask8) __U,
+             (__v4sf) __A,
+             (__v4sf) _mm_setzero_ps ());
 }
 
 static __inline__ __m256 __DEFAULT_FN_ATTRS
 _mm256_mask_mov_ps (__m256 __W, __mmask8 __U, __m256 __A)
 {
-  return (__m256) __builtin_ia32_movaps256_mask ((__v8sf) __A,
-             (__v8sf) __W,
-             (__mmask8) __U);
+  return (__m256) __builtin_ia32_selectps_256 ((__mmask8) __U,
+             (__v8sf) __A,
+             (__v8sf) __W);
 }
 
 static __inline__ __m256 __DEFAULT_FN_ATTRS
 _mm256_maskz_mov_ps (__mmask8 __U, __m256 __A)
 {
-  return (__m256) __builtin_ia32_movaps256_mask ((__v8sf) __A,
-             (__v8sf)
-             _mm256_setzero_ps (),
-             (__mmask8) __U);
+  return (__m256) __builtin_ia32_selectps_256 ((__mmask8) __U,
+             (__v8sf) __A,
+             (__v8sf) _mm256_setzero_ps ());
 }
 
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_mask_cvtph_ps (__m128 __W, __mmask8 __U, __m128i __A)
 {
   return (__m128) __builtin_ia32_vcvtph2ps_mask ((__v8hi) __A,
              (__v4sf) __W,
              (__mmask8) __U);

test/CodeGen/avx512bw-builtins.c

 __m512i test_mm512_maskz_mullo_epi16 (__mmask32 __U, __m512i __A, __m512i __B) {
   //CHECK-LABEL: @test_mm512_maskz_mullo_epi16
   //CHECK: @llvm.x86.avx512.mask.pmull.w.512
   return _mm512_maskz_mullo_epi16(__U, __A, __B);
 }
 
 __m512i test_mm512_mask_blend_epi8(__mmask64 __U, __m512i __A, __m512i __W) {
   // CHECK-LABEL: @test_mm512_mask_blend_epi8
-  // CHECK: @llvm.x86.avx512.mask.blend.b.512
+  // CHECK: select <64 x i1> %{{.*}}, <64 x i8> %{{.*}}, <64 x i8> %{{.*}}
   return _mm512_mask_blend_epi8(__U,__A,__W); 
 }
 __m512i test_mm512_mask_blend_epi16(__mmask32 __U, __m512i __A, __m512i __W) {
   // CHECK-LABEL: @test_mm512_mask_blend_epi16
-  // CHECK: @llvm.x86.avx512.mask.blend.w.512
+  // CHECK: select <32 x i1> %{{.*}}, <32 x i16> %{{.*}}, <32 x i16> %{{.*}}
   return _mm512_mask_blend_epi16(__U,__A,__W); 
 }
 __m512i test_mm512_abs_epi8(__m512i __A) {
   // CHECK-LABEL: @test_mm512_abs_epi8
   // CHECK: @llvm.x86.avx512.mask.pabs.b.512
   return _mm512_abs_epi8(__A); 
 }
 __m512i test_mm512_mask_abs_epi8(__m512i __W, __mmask64 __U, __m512i __A) {

 __m512i test_mm512_maskz_srli_epi16(__mmask32 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_maskz_srli_epi16
   // CHECK: @llvm.x86.avx512.mask.psrl.wi.512
   return _mm512_maskz_srli_epi16(__U, __A, 5); 
 }
 
 __m512i test_mm512_mask_mov_epi16(__m512i __W, __mmask32 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_mask_mov_epi16
-  // CHECK: @llvm.x86.avx512.mask.movu.w.512
+  // CHECK: select <32 x i1> %{{.*}}, <32 x i16> %{{.*}}, <32 x i16> %{{.*}}
   return _mm512_mask_mov_epi16(__W, __U, __A); 
 }
 
 __m512i test_mm512_maskz_mov_epi16(__mmask32 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_maskz_mov_epi16
-  // CHECK: @llvm.x86.avx512.mask.movu.w.512
+  // CHECK: select <32 x i1> %{{.*}}, <32 x i16> %{{.*}}, <32 x i16> %{{.*}}
   return _mm512_maskz_mov_epi16(__U, __A); 
 }
 
 __m512i test_mm512_mask_mov_epi8(__m512i __W, __mmask64 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_mask_mov_epi8
-  // CHECK: @llvm.x86.avx512.mask.movu.b.512
+  // CHECK: select <64 x i1> %{{.*}}, <64 x i8> %{{.*}}, <64 x i8> %{{.*}}
   return _mm512_mask_mov_epi8(__W, __U, __A); 
 }
 
 __m512i test_mm512_maskz_mov_epi8(__mmask64 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_maskz_mov_epi8
-  // CHECK: @llvm.x86.avx512.mask.movu.b.512
+  // CHECK: select <64 x i1> %{{.*}}, <64 x i8> %{{.*}}, <64 x i8> %{{.*}}
   return _mm512_maskz_mov_epi8(__U, __A); 
 }
 
 __m512i test_mm512_mask_set1_epi8(__m512i __O, __mmask64 __M, char __A) {
   // CHECK-LABEL: @test_mm512_mask_set1_epi8
   // CHECK: @llvm.x86.avx512.mask.pbroadcast.b.gpr.512
   return _mm512_mask_set1_epi8(__O, __M, __A); 
 }

test/CodeGen/avx512f-builtins.c

 __m512i test_mm512_maskz_load_epi32(__mmask16 __U, void const *__P) {
   // CHECK-LABEL: @test_mm512_maskz_load_epi32
   // CHECK: @llvm.masked.load.v16i32(<16 x i32>* %{{.*}}, i32 64, <16 x i1> %{{.*}}, <16 x i32> %{{.*}})
   return _mm512_maskz_load_epi32(__U, __P); 
 }
 
 __m512i test_mm512_mask_mov_epi32(__m512i __W, __mmask16 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_mask_mov_epi32
-  // CHECK: @llvm.x86.avx512.mask.mov
+  // CHECK: select <16 x i1> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> %{{.*}}
   return _mm512_mask_mov_epi32(__W, __U, __A); 
 }
 
 __m512i test_mm512_maskz_mov_epi32(__mmask16 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_maskz_mov_epi32
-  // CHECK: @llvm.x86.avx512.mask.mov
+  // CHECK: select <16 x i1> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> %{{.*}}
   return _mm512_maskz_mov_epi32(__U, __A); 
 }
 
 __m512i test_mm512_mask_mov_epi64(__m512i __W, __mmask8 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_mask_mov_epi64
-  // CHECK: @llvm.x86.avx512.mask.mov
+  // CHECK: select <8 x i1> %{{.*}}, <8 x i64> %{{.*}}, <8 x i64> %{{.*}}
   return _mm512_mask_mov_epi64(__W, __U, __A); 
 }
 
 __m512i test_mm512_maskz_mov_epi64(__mmask8 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_maskz_mov_epi64
-  // CHECK: @llvm.x86.avx512.mask.mov
+  // CHECK: select <8 x i1> %{{.*}}, <8 x i64> %{{.*}}, <8 x i64> %{{.*}}
   return _mm512_maskz_mov_epi64(__U, __A); 
 }
 
 __m512i test_mm512_mask_load_epi64(__m512i __W, __mmask8 __U, void const *__P) {
   // CHECK-LABEL: @test_mm512_mask_load_epi64
   // CHECK: @llvm.masked.load.v8i64(<8 x i64>* %{{.*}}, i32 64, <8 x i1> %{{.*}}, <8 x i64> %{{.*}})
   return _mm512_mask_load_epi64(__W, __U, __P); 
 }

 
 __m512d test_mm512_maskz_cvtps_pd(__mmask8 __U, __m256 __A) {
   // CHECK-LABEL: @test_mm512_maskz_cvtps_pd
   // CHECK: @llvm.x86.avx512.mask.cvtps2pd.512
   return _mm512_maskz_cvtps_pd(__U, __A); 
 }
 __m512d test_mm512_mask_mov_pd(__m512d __W, __mmask8 __U, __m512d __A) {
   // CHECK-LABEL: @test_mm512_mask_mov_pd
-  // CHECK: @llvm.x86.avx512.mask.mova.pd.512
+  // CHECK: select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
   return _mm512_mask_mov_pd(__W, __U, __A); 
 }
 
 __m512d test_mm512_maskz_mov_pd(__mmask8 __U, __m512d __A) {
   // CHECK-LABEL: @test_mm512_maskz_mov_pd
-  // CHECK: @llvm.x86.avx512.mask.mova.pd.512
+  // CHECK: select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
   return _mm512_maskz_mov_pd(__U, __A); 
 }
 
 __m512 test_mm512_mask_mov_ps(__m512 __W, __mmask16 __U, __m512 __A) {
   // CHECK-LABEL: @test_mm512_mask_mov_ps
-  // CHECK: @llvm.x86.avx512.mask.mova.ps.512
+  // CHECK: select <16 x i1> %{{.*}}, <16 x float> %{{.*}}, <16 x float> %{{.*}}
   return _mm512_mask_mov_ps(__W, __U, __A); 
 }
 
 __m512 test_mm512_maskz_mov_ps(__mmask16 __U, __m512 __A) {
   // CHECK-LABEL: @test_mm512_maskz_mov_ps
-  // CHECK: @llvm.x86.avx512.mask.mova.ps.512
+  // CHECK: select <16 x i1> %{{.*}}, <16 x float> %{{.*}}, <16 x float> %{{.*}}
   return _mm512_maskz_mov_ps(__U, __A); 
 }
 
 void test_mm512_mask_compressstoreu_pd(void *__P, __mmask8 __U, __m512d __A) {
   // CHECK-LABEL: @test_mm512_mask_compressstoreu_pd
   // CHECK: @llvm.x86.avx512.mask.compress.store.pd.512
   return _mm512_mask_compressstoreu_pd(__P, __U, __A); 
 }

 }
 
 __m512d test_mm512_setzero_pd()
 {
   // CHECK-LABEL: @test_mm512_setzero_pd
   // CHECK: zeroinitializer
   return _mm512_setzero_pd();
 }
 No newline at end of file

test/CodeGen/avx512vl-builtins.c

 }
 __m256 test_mm256_maskz_add_ps(__mmask16 __U, __m256 __A, __m256 __B) {
   // CHECK-LABEL: @test_mm256_maskz_add_ps
   // CHECK: @llvm.x86.avx512.mask.add.ps.256
   return _mm256_maskz_add_ps(__U,__A,__B); 
 }
 __m128i test_mm_mask_blend_epi32(__mmask8 __U, __m128i __A, __m128i __W) {
   // CHECK-LABEL: @test_mm_mask_blend_epi32
-  // CHECK: @llvm.x86.avx512.mask.blend.d.128
+  // CHECK: select <4 x i1> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> %{{.*}}
   return _mm_mask_blend_epi32(__U,__A,__W); 
 }
 __m256i test_mm256_mask_blend_epi32(__mmask8 __U, __m256i __A, __m256i __W) {
   // CHECK-LABEL: @test_mm256_mask_blend_epi32
-  // CHECK: @llvm.x86.avx512.mask.blend.d.256
+  // CHECK: select <8 x i1> %{{.*}}, <8 x i32> %{{.*}}, <8 x i32> %{{.*}}
   return _mm256_mask_blend_epi32(__U,__A,__W); 
 }
 __m128d test_mm_mask_blend_pd(__mmask8 __U, __m128d __A, __m128d __W) {
   // CHECK-LABEL: @test_mm_mask_blend_pd
-  // CHECK: @llvm.x86.avx512.mask.blend.pd.128
+  // CHECK: select <2 x i1> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}}
   return _mm_mask_blend_pd(__U,__A,__W); 
 }
 __m256d test_mm256_mask_blend_pd(__mmask8 __U, __m256d __A, __m256d __W) {
   // CHECK-LABEL: @test_mm256_mask_blend_pd
-  // CHECK: @llvm.x86.avx512.mask.blend.pd.256
+  // CHECK: select <4 x i1> %{{.*}}, <4 x double> %{{.*}}, <4 x double> %{{.*}}
   return _mm256_mask_blend_pd(__U,__A,__W); 
 }
 __m128 test_mm_mask_blend_ps(__mmask8 __U, __m128 __A, __m128 __W) {
   // CHECK-LABEL: @test_mm_mask_blend_ps
-  // CHECK: @llvm.x86.avx512.mask.blend.ps.128
+  // CHECK: select <4 x i1> %{{.*}}, <4 x float> %{{.*}}, <4 x float> %{{.*}}
   return _mm_mask_blend_ps(__U,__A,__W); 
 }
 __m256 test_mm256_mask_blend_ps(__mmask8 __U, __m256 __A, __m256 __W) {
   // CHECK-LABEL: @test_mm256_mask_blend_ps
-  // CHECK: @llvm.x86.avx512.mask.blend.ps.256
+  // CHECK: select <8 x i1> %{{.*}}, <8 x float> %{{.*}}, <8 x float> %{{.*}}
   return _mm256_mask_blend_ps(__U,__A,__W); 
 }
 __m128i test_mm_mask_blend_epi64(__mmask8 __U, __m128i __A, __m128i __W) {
   // CHECK-LABEL: @test_mm_mask_blend_epi64
-  // CHECK: @llvm.x86.avx512.mask.blend.q.128
+  // CHECK: select <2 x i1> %{{.*}}, <2 x i64> %{{.*}}, <2 x i64> %{{.*}}
   return _mm_mask_blend_epi64(__U,__A,__W); 
 }
 __m256i test_mm256_mask_blend_epi64(__mmask8 __U, __m256i __A, __m256i __W) {
   // CHECK-LABEL: @test_mm256_mask_blend_epi64
-  // CHECK: @llvm.x86.avx512.mask.blend.q.256
+  // CHECK: select <4 x i1> %{{.*}}, <4 x i64> %{{.*}}, <4 x i64> %{{.*}}
   return _mm256_mask_blend_epi64(__U,__A,__W); 
 }
 __m128d test_mm_mask_compress_pd(__m128d __W, __mmask8 __U, __m128d __A) {
   // CHECK-LABEL: @test_mm_mask_compress_pd
   // CHECK: @llvm.x86.avx512.mask.compress.pd.128
   return _mm_mask_compress_pd(__W,__U,__A); 
 }
 __m128d test_mm_maskz_compress_pd(__mmask8 __U, __m128d __A) {

 void test_mm256_mask_store_epi32(void *__P, __mmask8 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_mask_store_epi32
   // CHECK: @llvm.masked.store.v8i32(<8 x i32> %{{.*}}, <8 x i32>* %{{.}}, i32 32, <8 x i1> %{{.*}})
   return _mm256_mask_store_epi32(__P, __U, __A); 
 }
 
 __m128i test_mm_mask_mov_epi32(__m128i __W, __mmask8 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_mask_mov_epi32
-  // CHECK: @llvm.x86.avx512.mask.mov
+  // CHECK: select <4 x i1> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> %{{.*}}
   return _mm_mask_mov_epi32(__W, __U, __A); 
 }
 
 __m128i test_mm_maskz_mov_epi32(__mmask8 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_maskz_mov_epi32
-  // CHECK: @llvm.x86.avx512.mask.mov
+  // CHECK: select <4 x i1> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> %{{.*}}
   return _mm_maskz_mov_epi32(__U, __A); 
 }
 
 __m256i test_mm256_mask_mov_epi32(__m256i __W, __mmask8 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_mask_mov_epi32
-  // CHECK: @llvm.x86.avx512.mask.mov
+  // CHECK: select <8 x i1> %{{.*}}, <8 x i32> %{{.*}}, <8 x i32> %{{.*}}
   return _mm256_mask_mov_epi32(__W, __U, __A); 
 }
 
 __m256i test_mm256_maskz_mov_epi32(__mmask8 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_maskz_mov_epi32
-  // CHECK: @llvm.x86.avx512.mask.mov
+  // CHECK: select <8 x i1> %{{.*}}, <8 x i32> %{{.*}}, <8 x i32> %{{.*}}
   return _mm256_maskz_mov_epi32(__U, __A); 
 }
 
 __m128i test_mm_mask_mov_epi64(__m128i __W, __mmask8 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_mask_mov_epi64
-  // CHECK: @llvm.x86.avx512.mask.mov
+  // CHECK: select <2 x i1> %{{.*}}, <2 x i64> %{{.*}}, <2 x i64> %{{.*}}
   return _mm_mask_mov_epi64(__W, __U, __A); 
 }
 
 __m128i test_mm_maskz_mov_epi64(__mmask8 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_maskz_mov_epi64
-  // CHECK: @llvm.x86.avx512.mask.mov
+  // CHECK: select <2 x i1> %{{.*}}, <2 x i64> %{{.*}}, <2 x i64> %{{.*}}
   return _mm_maskz_mov_epi64(__U, __A); 
 }
 
 __m256i test_mm256_mask_mov_epi64(__m256i __W, __mmask8 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_mask_mov_epi64
-  // CHECK: @llvm.x86.avx512.mask.mov
+  // CHECK: select <4 x i1> %{{.*}}, <4 x i64> %{{.*}}, <4 x i64> %{{.*}}
   return _mm256_mask_mov_epi64(__W, __U, __A); 
 }
 
 __m256i test_mm256_maskz_mov_epi64(__mmask8 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_maskz_mov_epi64
-  // CHECK: @llvm.x86.avx512.mask.mov
+  // CHECK: select <4 x i1> %{{.*}}, <4 x i64> %{{.*}}, <4 x i64> %{{.*}}
   return _mm256_maskz_mov_epi64(__U, __A); 
 }
 
 __m128i test_mm_mask_load_epi32(__m128i __W, __mmask8 __U, void const *__P) {
   // CHECK-LABEL: @test_mm_mask_load_epi32
   // CHECK: @llvm.masked.load.v4i32(<4 x i32>* %{{.*}}, i32 16, <4 x i1> %{{.*}}, <4 x i32> %{{.*}})
   return _mm_mask_load_epi32(__W, __U, __P); 
 }

 __m256i test_mm256_maskz_shuffle_epi32(__mmask8 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_maskz_shuffle_epi32
   // CHECK: @llvm.x86.avx512.mask.pshuf.d.256
   return _mm256_maskz_shuffle_epi32(__U, __A, 2); 
 }
 
 __m128d test_mm_mask_mov_pd(__m128d __W, __mmask8 __U, __m128d __A) {
   // CHECK-LABEL: @test_mm_mask_mov_pd
-  // CHECK: @llvm.x86.avx512.mask.mova.pd.128
+  // CHECK: select <2 x i1> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}}
   return _mm_mask_mov_pd(__W, __U, __A); 
 }
 
 __m128d test_mm_maskz_mov_pd(__mmask8 __U, __m128d __A) {
   // CHECK-LABEL: @test_mm_maskz_mov_pd
-  // CHECK: @llvm.x86.avx512.mask.mova.pd.128
+  // CHECK: select <2 x i1> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}}
   return _mm_maskz_mov_pd(__U, __A); 
 }
 
 __m256d test_mm256_mask_mov_pd(__m256d __W, __mmask8 __U, __m256d __A) {
   // CHECK-LABEL: @test_mm256_mask_mov_pd
-  // CHECK: @llvm.x86.avx512.mask.mova.pd.256
+  // CHECK: select <4 x i1> %{{.*}}, <4 x double> %{{.*}}, <4 x double> %{{.*}}
   return _mm256_mask_mov_pd(__W, __U, __A); 
 }
 
 __m256d test_mm256_maskz_mov_pd(__mmask8 __U, __m256d __A) {
   // CHECK-LABEL: @test_mm256_maskz_mov_pd
-  // CHECK: @llvm.x86.avx512.mask.mova.pd.256
+  // CHECK: select <4 x i1> %{{.*}}, <4 x double> %{{.*}}, <4 x double> %{{.*}}
   return _mm256_maskz_mov_pd(__U, __A); 
 }
 
 __m128 test_mm_mask_mov_ps(__m128 __W, __mmask8 __U, __m128 __A) {
   // CHECK-LABEL: @test_mm_mask_mov_ps
-  // CHECK: @llvm.x86.avx512.mask.mova.ps.128
+  // CHECK: select <4 x i1> %{{.*}}, <4 x float> %{{.*}}, <4 x float> %{{.*}}
   return _mm_mask_mov_ps(__W, __U, __A); 
 }
 
 __m128 test_mm_maskz_mov_ps(__mmask8 __U, __m128 __A) {
   // CHECK-LABEL: @test_mm_maskz_mov_ps
-  // CHECK: @llvm.x86.avx512.mask.mova.ps.128
+  // CHECK: select <4 x i1> %{{.*}}, <4 x float> %{{.*}}, <4 x float> %{{.*}}
   return _mm_maskz_mov_ps(__U, __A); 
 }
 
 __m256 test_mm256_mask_mov_ps(__m256 __W, __mmask8 __U, __m256 __A) {
   // CHECK-LABEL: @test_mm256_mask_mov_ps
-  // CHECK: @llvm.x86.avx512.mask.mova.ps.256
+  // CHECK: select <8 x i1> %{{.*}}, <8 x float> %{{.*}}, <8 x float> %{{.*}}
   return _mm256_mask_mov_ps(__W, __U, __A); 
 }
 
 __m256 test_mm256_maskz_mov_ps(__mmask8 __U, __m256 __A) {
   // CHECK-LABEL: @test_mm256_maskz_mov_ps
-  // CHECK: @llvm.x86.avx512.mask.mova.ps.256
+  // CHECK: select <8 x i1> %{{.*}}, <8 x float> %{{.*}}, <8 x float> %{{.*}}
   return _mm256_maskz_mov_ps(__U, __A); 
 }
 
 __m128 test_mm_mask_cvtph_ps(__m128 __W, __mmask8 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_mask_cvtph_ps
   // CHECK: @llvm.x86.avx512.mask.vcvtph2ps.128
   return _mm_mask_cvtph_ps(__W, __U, __A); 
 }

test/CodeGen/avx512vlbw-builtins.c

__m128i test_mm_maskz_mullo_epi16 (__mmask8 __U, __m128i __A, __m128i __B) {
   //CHECK-LABEL: @test_mm_maskz_mullo_epi16
   //CHECK: @llvm.x86.avx512.mask.pmull.w.128
   return _mm_maskz_mullo_epi16(__U , __A, __B);
 }
 
 
 __m128i test_mm_mask_blend_epi8(__mmask16 __U, __m128i __A, __m128i __W) {
   // CHECK-LABEL: @test_mm_mask_blend_epi8
-  // CHECK: @llvm.x86.avx512.mask.blend.b.128
+  // CHECK: select <16 x i1> %{{.*}}, <16 x i8> %{{.*}}, <16 x i8> %{{.*}}
   return _mm_mask_blend_epi8(__U,__A,__W); 
 }
 __m256i test_mm256_mask_blend_epi8(__mmask32 __U, __m256i __A, __m256i __W) {
   // CHECK-LABEL: @test_mm256_mask_blend_epi8
-  // CHECK: @llvm.x86.avx512.mask.blend.b.256
+  // CHECK: select <32 x i1> %{{.*}}, <32 x i8> %{{.*}}, <32 x i8> %{{.*}}
   return _mm256_mask_blend_epi8(__U,__A,__W); 
 }
 
 __m128i test_mm_mask_blend_epi16(__mmask8 __U, __m128i __A, __m128i __W) {
   // CHECK-LABEL: @test_mm_mask_blend_epi16
-  // CHECK: @llvm.x86.avx512.mask.blend.w.128
+  // CHECK: select <8 x i1> %{{.*}}, <8 x i16> %{{.*}}, <8 x i16> %{{.*}}
   return _mm_mask_blend_epi16(__U,__A,__W); 
 }
 
 __m256i test_mm256_mask_blend_epi16(__mmask16 __U, __m256i __A, __m256i __W) {
   // CHECK-LABEL: @test_mm256_mask_blend_epi16
-  // CHECK: @llvm.x86.avx512.mask.blend.w.256
+  // CHECK: select <16 x i1> %{{.*}}, <16 x i16> %{{.*}}, <16 x i16> %{{.*}}
   return _mm256_mask_blend_epi16(__U,__A,__W); 
 }
 
 __m128i test_mm_mask_abs_epi8(__m128i __W, __mmask16 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_mask_abs_epi8
   // CHECK: @llvm.x86.avx512.mask.pabs.b.128
   return _mm_mask_abs_epi8(__W,__U,__A); 
 }

 __m256i test_mm256_maskz_slli_epi16(__mmask16 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_maskz_slli_epi16
   // CHECK: @llvm.x86.avx512.mask.psll.wi.256
   return _mm256_maskz_slli_epi16(__U, __A, 5); 
 }
 
 __m128i test_mm_mask_mov_epi16(__m128i __W, __mmask8 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_mask_mov_epi16
-  // CHECK: @llvm.x86.avx512.mask.movu.w.128
+  // CHECK: select <8 x i1> %{{.*}}, <8 x i16> %{{.*}}, <8 x i16> %{{.*}}
   return _mm_mask_mov_epi16(__W, __U, __A); 
 }
 
 __m128i test_mm_maskz_mov_epi16(__mmask8 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_maskz_mov_epi16
-  // CHECK: @llvm.x86.avx512.mask.movu.w.128
+  // CHECK: select <8 x i1> %{{.*}}, <8 x i16> %{{.*}}, <8 x i16> %{{.*}}
   return _mm_maskz_mov_epi16(__U, __A); 
 }
 
 __m256i test_mm256_mask_mov_epi16(__m256i __W, __mmask16 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_mask_mov_epi16
-  // CHECK: @llvm.x86.avx512.mask.movu.w.256
+  // CHECK: select <16 x i1> %{{.*}}, <16 x i16> %{{.*}}, <16 x i16> %{{.*}}
   return _mm256_mask_mov_epi16(__W, __U, __A); 
 }
 
 __m256i test_mm256_maskz_mov_epi16(__mmask16 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_maskz_mov_epi16
-  // CHECK: @llvm.x86.avx512.mask.movu.w.256
+  // CHECK: select <16 x i1> %{{.*}}, <16 x i16> %{{.*}}, <16 x i16> %{{.*}}
   return _mm256_maskz_mov_epi16(__U, __A); 
 }
 
 __m128i test_mm_mask_mov_epi8(__m128i __W, __mmask16 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_mask_mov_epi8
-  // CHECK: @llvm.x86.avx512.mask.movu.b.128
+  // CHECK: select <16 x i1> %{{.*}}, <16 x i8> %{{.*}}, <16 x i8> %{{.*}}
   return _mm_mask_mov_epi8(__W, __U, __A); 
 }
 
 __m128i test_mm_maskz_mov_epi8(__mmask16 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_maskz_mov_epi8
-  // CHECK: @llvm.x86.avx512.mask.movu.b.128
+  // CHECK: select <16 x i1> %{{.*}}, <16 x i8> %{{.*}}, <16 x i8> %{{.*}}
   return _mm_maskz_mov_epi8(__U, __A); 
 }
 
 __m256i test_mm256_mask_mov_epi8(__m256i __W, __mmask32 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_mask_mov_epi8
-  // CHECK: @llvm.x86.avx512.mask.movu.b.256
+  // CHECK: select <32 x i1> %{{.*}}, <32 x i8> %{{.*}}, <32 x i8> %{{.*}}
   return _mm256_mask_mov_epi8(__W, __U, __A); 
 }
 
 __m256i test_mm256_maskz_mov_epi8(__mmask32 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_maskz_mov_epi8
-  // CHECK: @llvm.x86.avx512.mask.movu.b.256
+  // CHECK: select <32 x i1> %{{.*}}, <32 x i8> %{{.*}}, <32 x i8> %{{.*}}
   return _mm256_maskz_mov_epi8(__U, __A); 
 }
 
 __m128i test_mm_mask_loadu_epi16(__m128i __W, __mmask8 __U, void const *__P) {
   // CHECK-LABEL: @test_mm_mask_loadu_epi16
   // CHECK: @llvm.masked.load.v8i16(<8 x i16>* %{{.*}}, i32 1, <8 x i1> %{{.*}}, <8 x i16> %{{.*}})
   return _mm_mask_loadu_epi16(__W, __U, __P); 
 }