[X86] VP2INTERSECT clang

[xiangzhangllvm]

Summary

Support intel AVX512 VP2INTERSECT instructions in clang

[craig.topper]

lib/Basic/Targets/X86.cpp

518

This patch needs to be rebased. I modified the formatting here earlier to fix the weird indentation.

[xiangzhangllvm]

rebase

[craig.topper]

lib/Headers/avx512vlvp2intersectintrin.h

40

Can you add doxygen comments for the new intrinsics? @RKSimon has been asking for it on other reviews. I forgot to say something in our internal review.

[xiangzhangllvm]

lib/Headers/avx512vlvp2intersectintrin.h

40

OK! But I really find many ntrinsicxxx.h have no doxygen comments, Is it like this format: ?

/// Rounds up each element of the 128-bit vector of [4 x float] to an
///    integer and returns the rounded values in a 128-bit vector of
///    [4 x float].
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128 _mm_ceil_ps(__m128 X);
/// \endcode
///
/// This intrinsic corresponds to the <c> VROUNDPS / ROUNDPS </c> instruction.
///
/// \param X
///    A 128-bit vector of [4 x float] values to be rounded up.
/// \returns A 128-bit vector of [4 x float] containing the rounded values.
#define _mm_ceil_ps(X)       _mm_round_ps((X), _MM_FROUND_CEIL)

[craig.topper]

lib/Headers/avx512vlvp2intersectintrin.h

40

Yeah that looks right. Someone was kind enough to provide comments for some of the header files a few years ago, but it wasn't complete. I know avx512* especially are missing comments.

[RKSimon]

lib/Headers/avx512vlvp2intersectintrin.h

40

The hope is that these will get added as time goes on - usually it just requires a copy + paste of existing comments and tweak for ymm/zmm/whatever.

New intrinsics should always include documentation since the person adding it is the most likely to be able to correct describe it.

[xiangzhangllvm]

lib/Headers/avx512vlvp2intersectintrin.h

40

Yes, That make sense!

[xiangzhangllvm]

rebase

[xiangzhangllvm]

rebase

[xiangzhangllvm]

Hi Dear friends, could you help merge this patch? Thank you very much!

[craig.topper]

lib/Headers/avx512vlvp2intersectintrin.h

40

This doesn't really say anything about what's in the mask registers. I don' think any of the MODRM encoding details are relevant here. I think a better description would be what's in the table in the ISE manual. "Store, in an even/odd pair of mask registers,
the indicators of the locations of value
matches between dwords in
xmm3/m128/m32bcst and xmm2."

47

Drop the "AVX512-". Add the proper D/Q suffix. This should be the instruction mnemonic

54

ponit?

[xiangzhangllvm]

Done, Thank you very much!

[craig.topper]

LGTM

docs/ClangCommandLineReference.rst

 .. option:: -mavx512vbmi, -mno-avx512vbmi
 
 .. option:: -mavx512vbmi2, -mno-avx512vbmi2
 
 .. option:: -mavx512vl, -mno-avx512vl
 
 .. option:: -mavx512vnni, -mno-avx512vnni
 
+.. option:: -mavx512vp2intersect, -mno-avx512vp2intersect
+
 .. option:: -mavx512vpopcntdq, -mno-avx512vpopcntdq
 
 .. option:: -mbmi, -mno-bmi
 
 .. option:: -mbmi2, -mno-bmi2
 
 .. option:: -mcldemote, -mno-cldemote
 

include/clang/Basic/BuiltinsX86.def

 TARGET_BUILTIN(__builtin_ia32_cvtne2ps2bf16_256, "V16sV8fV8f", "ncV:256:", "avx512bf16,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_cvtne2ps2bf16_512, "V32sV16fV16f", "ncV:512:", "avx512bf16")
 TARGET_BUILTIN(__builtin_ia32_cvtneps2bf16_128_mask, "V8sV4fV8sUc", "ncV:128:", "avx512bf16,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_cvtneps2bf16_256_mask, "V8sV8fV8sUc", "ncV:256:", "avx512bf16,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_cvtneps2bf16_512_mask, "V16sV16fV16sUs", "ncV:512:", "avx512bf16")
 TARGET_BUILTIN(__builtin_ia32_dpbf16ps_128, "V4fV4fV4iV4i", "ncV:128:", "avx512bf16,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_dpbf16ps_256, "V8fV8fV8iV8i", "ncV:256:", "avx512bf16,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_dpbf16ps_512, "V16fV16fV16iV16i", "ncV:512:", "avx512bf16")
+TARGET_BUILTIN(__builtin_ia32_vp2intersect_q_512, "vV8LLiV8LLiUc*Uc*", "nV:512:", "avx512vp2intersect")
+TARGET_BUILTIN(__builtin_ia32_vp2intersect_q_256, "vV4LLiV4LLiUc*Uc*", "nV:256:", "avx512vp2intersect,avx512vl")
+TARGET_BUILTIN(__builtin_ia32_vp2intersect_q_128, "vV2LLiV2LLiUc*Uc*", "nV:128:", "avx512vp2intersect,avx512vl")
+TARGET_BUILTIN(__builtin_ia32_vp2intersect_d_512, "vV16iV16iUs*Us*", "nV:512:", "avx512vp2intersect")
+TARGET_BUILTIN(__builtin_ia32_vp2intersect_d_256, "vV8iV8iUc*Uc*", "nV:256:", "avx512vp2intersect,avx512vl")
+TARGET_BUILTIN(__builtin_ia32_vp2intersect_d_128, "vV4iV4iUc*Uc*", "nV:128:", "avx512vp2intersect,avx512vl")
 
 // generic select intrinsics
 TARGET_BUILTIN(__builtin_ia32_selectb_128, "V16cUsV16cV16c", "ncV:128:", "avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_selectb_256, "V32cUiV32cV32c", "ncV:256:", "avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_selectb_512, "V64cULLiV64cV64c", "ncV:512:", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_selectw_128, "V8sUcV8sV8s", "ncV:128:", "avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_selectw_256, "V16sUsV16sV16s", "ncV:256:", "avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_selectw_512, "V32sUiV32sV32s", "ncV:512:", "avx512bw")

include/clang/Driver/Options.td

 def mavx512vbmi2 : Flag<["-"], "mavx512vbmi2">, Group<m_x86_Features_Group>;
 def mno_avx512vbmi2 : Flag<["-"], "mno-avx512vbmi2">, Group<m_x86_Features_Group>;
 def mavx512vl : Flag<["-"], "mavx512vl">, Group<m_x86_Features_Group>;
 def mno_avx512vl : Flag<["-"], "mno-avx512vl">, Group<m_x86_Features_Group>;
 def mavx512vnni : Flag<["-"], "mavx512vnni">, Group<m_x86_Features_Group>;
 def mno_avx512vnni : Flag<["-"], "mno-avx512vnni">, Group<m_x86_Features_Group>;
 def mavx512vpopcntdq : Flag<["-"], "mavx512vpopcntdq">, Group<m_x86_Features_Group>;
 def mno_avx512vpopcntdq : Flag<["-"], "mno-avx512vpopcntdq">, Group<m_x86_Features_Group>;
+def mavx512vp2intersect : Flag<["-"], "mavx512vp2intersect">, Group<m_x86_Features_Group>;
+def mno_avx512vp2intersect : Flag<["-"], "mno-avx512vp2intersect">, Group<m_x86_Features_Group>;
 def madx : Flag<["-"], "madx">, Group<m_x86_Features_Group>;
 def mno_adx : Flag<["-"], "mno-adx">, Group<m_x86_Features_Group>;
 def maes : Flag<["-"], "maes">, Group<m_x86_Features_Group>;
 def mno_aes : Flag<["-"], "mno-aes">, Group<m_x86_Features_Group>;
 def mbmi : Flag<["-"], "mbmi">, Group<m_x86_Features_Group>;
 def mno_bmi : Flag<["-"], "mno-bmi">, Group<m_x86_Features_Group>;
 def mbmi2 : Flag<["-"], "mbmi2">, Group<m_x86_Features_Group>;
 def mno_bmi2 : Flag<["-"], "mno-bmi2">, Group<m_x86_Features_Group>;

lib/Basic/Targets/X86.h

class LLVM_LIBRARY_VISIBILITY X86TargetInfo : public TargetInfo {
   bool HasAVX512PF = false;
   bool HasAVX512DQ = false;
   bool HasAVX512BITALG = false;
   bool HasAVX512BW = false;
   bool HasAVX512VL = false;
   bool HasAVX512VBMI = false;
   bool HasAVX512VBMI2 = false;
   bool HasAVX512IFMA = false;
+  bool HasAVX512VP2INTERSECT = false;
   bool HasSHA = false;
   bool HasMPX = false;
   bool HasSHSTK = false;
   bool HasSGX = false;
   bool HasCX8 = false;
   bool HasCX16 = false;
   bool HasFXSR = false;
   bool HasXSAVE = false;

lib/Basic/Targets/X86.cpp

case AVX:
     Features["fma"] = Features["avx"] = Features["f16c"] = Features["xsave"] =
         Features["xsaveopt"] = Features["vaes"] = Features["vpclmulqdq"] = false;
     setXOPLevel(Features, FMA4, false);
     LLVM_FALLTHROUGH;
   case AVX2:
     Features["avx2"] = false;
     LLVM_FALLTHROUGH;
   case AVX512F:
     Features["avx512f"] = Features["avx512cd"] = Features["avx512er"] =
         Features["avx512pf"] = Features["avx512dq"] = Features["avx512bw"] =
             Features["avx512vl"] = Features["avx512vbmi"] =
                 Features["avx512ifma"] = Features["avx512vpopcntdq"] =
                     Features["avx512bitalg"] = Features["avx512vnni"] =
                         Features["avx512vbmi2"] = false;
                         Features["avx512bf16"] = false;
+                        Features["avx512vp2intersect"] = false;
     break;
   }
 }
 
 void X86TargetInfo::setMMXLevel(llvm::StringMap<bool> &Features,
                                 MMX3DNowEnum Level, bool Enabled) {
   if (Enabled) {
     switch (Level) {
void X86TargetInfo::setFeatureEnabledImpl(llvm::StringMap<bool> &Features,
   } else if (Name == "avx2") {
     setSSELevel(Features, AVX2, Enabled);
   } else if (Name == "avx512f") {
     setSSELevel(Features, AVX512F, Enabled);
   } else if (Name == "avx512cd" || Name == "avx512er" || Name == "avx512pf" ||
              Name == "avx512dq" || Name == "avx512bw" || Name == "avx512vl" ||
              Name == "avx512vbmi" || Name == "avx512ifma" ||
              Name == "avx512vpopcntdq" || Name == "avx512bitalg" ||
-             Name == "avx512bf16" ||
+             Name == "avx512vp2intersect" || Name == "avx512bf16" ||
              Name == "avx512vnni" || Name == "avx512vbmi2") {
     if (Enabled)
       setSSELevel(Features, AVX512F, Enabled);
     // Enable BWI instruction if VBMI/VBMI2/BITALG is being enabled.
     if ((Name.startswith("avx512vbmi") || Name == "avx512bitalg") && Enabled)
       Features["avx512bw"] = true;
     if (Name == "avx512bf16" && Enabled)
       Features["avx512bw"] = true;
for (const auto &Feature : Features) {
     } else if (Feature == "+avx512vl") {
       HasAVX512VL = true;
     } else if (Feature == "+avx512vbmi") {
       HasAVX512VBMI = true;
     } else if (Feature == "+avx512vbmi2") {
       HasAVX512VBMI2 = true;
     } else if (Feature == "+avx512ifma") {
       HasAVX512IFMA = true;
+    } else if (Feature == "+avx512vp2intersect") {
+      HasAVX512VP2INTERSECT = true;
     } else if (Feature == "+sha") {
       HasSHA = true;
     } else if (Feature == "+mpx") {
       HasMPX = true;
     } else if (Feature == "+shstk") {
       HasSHSTK = true;
     } else if (Feature == "+movbe") {
       HasMOVBE = true;
void X86TargetInfo::getTargetDefines(const LangOptions &Opts,
   if (HasAVX512VL)
     Builder.defineMacro("__AVX512VL__");
   if (HasAVX512VBMI)
     Builder.defineMacro("__AVX512VBMI__");
   if (HasAVX512VBMI2)
     Builder.defineMacro("__AVX512VBMI2__");
   if (HasAVX512IFMA)
     Builder.defineMacro("__AVX512IFMA__");
-
+  if (HasAVX512VP2INTERSECT)
+    Builder.defineMacro("__AVX512VP2INTERSECT__");
   if (HasSHA)
     Builder.defineMacro("__SHA__");
 
   if (HasFXSR)
     Builder.defineMacro("__FXSR__");
   if (HasXSAVE)
     Builder.defineMacro("__XSAVE__");
   if (HasXSAVEOPT)
return llvm::StringSwitch<bool>(Name)
       .Case("avx512pf", true)
       .Case("avx512dq", true)
       .Case("avx512bitalg", true)
       .Case("avx512bw", true)
       .Case("avx512vl", true)
       .Case("avx512vbmi", true)
       .Case("avx512vbmi2", true)
       .Case("avx512ifma", true)
+      .Case("avx512vp2intersect", true)
       .Case("bmi", true)
       .Case("bmi2", true)
       .Case("cldemote", true)
       .Case("clflushopt", true)
       .Case("clwb", true)
       .Case("clzero", true)
       .Case("cx16", true)
       .Case("f16c", true)
return llvm::StringSwitch<bool>(Feature)
       .Case("avx512pf", HasAVX512PF)
       .Case("avx512dq", HasAVX512DQ)
       .Case("avx512bitalg", HasAVX512BITALG)
       .Case("avx512bw", HasAVX512BW)
       .Case("avx512vl", HasAVX512VL)
       .Case("avx512vbmi", HasAVX512VBMI)
       .Case("avx512vbmi2", HasAVX512VBMI2)
       .Case("avx512ifma", HasAVX512IFMA)
+      .Case("avx512vp2intersect", HasAVX512VP2INTERSECT)
       .Case("bmi", HasBMI)
       .Case("bmi2", HasBMI2)
       .Case("cldemote", HasCLDEMOTE)
       .Case("clflushopt", HasCLFLUSHOPT)
       .Case("clwb", HasCLWB)
       .Case("clzero", HasCLZERO)
       .Case("cx8", HasCX8)
       .Case("cx16", HasCX16)

lib/CodeGen/CGBuiltin.cpp

case X86::BI__builtin_ia32_fpclasspd512_mask:
       ID = Intrinsic::x86_avx512_fpclass_pd_512;
       break;
     }
 
     Value *Fpclass = Builder.CreateCall(CGM.getIntrinsic(ID), Ops);
     return EmitX86MaskedCompareResult(*this, Fpclass, NumElts, MaskIn);
   }
 
+  case X86::BI__builtin_ia32_vp2intersect_q_512:
+  case X86::BI__builtin_ia32_vp2intersect_q_256:
+  case X86::BI__builtin_ia32_vp2intersect_q_128:
+  case X86::BI__builtin_ia32_vp2intersect_d_512:
+  case X86::BI__builtin_ia32_vp2intersect_d_256:
+  case X86::BI__builtin_ia32_vp2intersect_d_128: {
+    unsigned NumElts = Ops[0]->getType()->getVectorNumElements();
+    Intrinsic::ID ID;
+
+    switch (BuiltinID) {
+    default: llvm_unreachable("Unsupported intrinsic!");
+    case X86::BI__builtin_ia32_vp2intersect_q_512:
+      ID = Intrinsic::x86_avx512_vp2intersect_q_512;
+      break;
+    case X86::BI__builtin_ia32_vp2intersect_q_256:
+      ID = Intrinsic::x86_avx512_vp2intersect_q_256;
+      break;
+    case X86::BI__builtin_ia32_vp2intersect_q_128:
+      ID = Intrinsic::x86_avx512_vp2intersect_q_128;
+      break;
+    case X86::BI__builtin_ia32_vp2intersect_d_512:
+      ID = Intrinsic::x86_avx512_vp2intersect_d_512;
+      break;
+    case X86::BI__builtin_ia32_vp2intersect_d_256:
+      ID = Intrinsic::x86_avx512_vp2intersect_d_256;
+      break;
+    case X86::BI__builtin_ia32_vp2intersect_d_128:
+      ID = Intrinsic::x86_avx512_vp2intersect_d_128;
+      break;
+    }
+
+    Value *Call = Builder.CreateCall(CGM.getIntrinsic(ID), {Ops[0], Ops[1]});
+    Value *Result = Builder.CreateExtractValue(Call, 0);
+    Result = EmitX86MaskedCompareResult(*this, Result, NumElts, nullptr);
+    Value *Store = Builder.CreateDefaultAlignedStore(Result, Ops[2]);
+
+    Result = Builder.CreateExtractValue(Call, 1);
+    Result = EmitX86MaskedCompareResult(*this, Result, NumElts, nullptr);
+    Store = Builder.CreateDefaultAlignedStore(Result, Ops[3]);
+    return Store;
+  }
+
   case X86::BI__builtin_ia32_vpmultishiftqb128:
   case X86::BI__builtin_ia32_vpmultishiftqb256:
   case X86::BI__builtin_ia32_vpmultishiftqb512: {
     Intrinsic::ID ID;
     switch (BuiltinID) {
     default: llvm_unreachable("Unsupported intrinsic!");
     case X86::BI__builtin_ia32_vpmultishiftqb128:
       ID = Intrinsic::x86_avx512_pmultishift_qb_128;

lib/Headers/CMakeLists.txt

set(files
   avx512vbmivlintrin.h
   avx512vbmi2intrin.h
   avx512vlvbmi2intrin.h
   avx512vlbf16intrin.h
   avx512vlbwintrin.h
   avx512vlcdintrin.h
   avx512vldqintrin.h
   avx512vlintrin.h
+  avx512vp2intersectintrin.h
+  avx512vlvp2intersectintrin.h
   avx512vpopcntdqvlintrin.h
   avx512vnniintrin.h
   avx512vlvnniintrin.h
   avxintrin.h
   bmi2intrin.h
   bmiintrin.h
   __clang_cuda_builtin_vars.h
   __clang_cuda_cmath.h

lib/Headers/avx512vlvp2intersectintrin.h

+/*===------ avx512vlvp2intersectintrin.h - VL VP2INTERSECT intrinsics ------===
+ *
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ *
+ *===-----------------------------------------------------------------------===
+ */
+#ifndef __IMMINTRIN_H
+#error "Never use <avx512vlvp2intersectintrin.h> directly; include <immintrin.h> instead."
+#endif
+
+#ifndef _AVX512VLVP2INTERSECT_H
+#define _AVX512VLVP2INTERSECT_H
+
+#define __DEFAULT_FN_ATTRS128 \
+  __attribute__((__always_inline__, __nodebug__,  __target__("avx512vl,avx512vp2intrin"), \
+                 __min_vector_width__(128)))
+
+#define __DEFAULT_FN_ATTRS256 \
+  __attribute__((__always_inline__, __nodebug__,  __target__("avx512vl,avx512vp2intrin"), \
+                 __min_vector_width__(256)))
+
+static __inline__ void __DEFAULT_FN_ATTRS256
+_mm256_2intersect_epi32(__m256i __a, __m256i __b, __mmask8 *__m0, __mmask8 *__m1) {
+  __builtin_ia32_vp2intersect_d_256((__v8si)__a, (__v8si)__b, __m0, __m1);
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS256
+_mm256_2intersect_epi64(__m256i __a, __m256i __b, __mmask8 *__m0, __mmask8 *__m1) {
+  __builtin_ia32_vp2intersect_q_256((__v4di)__a, (__v4di)__b, __m0, __m1);
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS128
+_mm_2intersect_epi32(__m128i __a, __m128i __b, __mmask8 *__m0, __mmask8 *__m1) {
+  __builtin_ia32_vp2intersect_d_128((__v4si)__a, (__v4si)__b, __m0, __m1);
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS128
+_mm_2intersect_epi64(__m128i __a, __m128i __b, __mmask8 *__m0, __mmask8 *__m1) {
+  __builtin_ia32_vp2intersect_q_128((__v2di)__a, (__v2di)__b, __m0, __m1);
+}
+
+#undef __DEFAULT_FN_ATTRS
+
+#endif

lib/Headers/avx512vp2intersectintrin.h

+/*===------- avx512vpintersectintrin.h - VP2INTERSECT intrinsics ------------===
+ *
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ *
+ *===-----------------------------------------------------------------------===
+ */
+#ifndef __IMMINTRIN_H
+#error "Never use <avx512vp2intersect.h> directly; include <immintrin.h> instead."
+#endif
+
+#ifndef _AVX512VP2INTERSECT_H
+#define _AVX512VP2INTERSECT_H
+
+#define __DEFAULT_FN_ATTRS \
+  __attribute__((__always_inline__, __nodebug__,  __target__("avx512vp2intrin"), \
+                 __min_vector_width__(512)))
+
+static __inline__ void __DEFAULT_FN_ATTRS
+_mm512_2intersect_epi32(__m512i __a, __m512i __b, __mmask16 *__m0, __mmask16 *__m1) {
+  __builtin_ia32_vp2intersect_d_512((__v16si)__a, (__v16si)__b, __m0, __m1);
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS
+_mm512_2intersect_epi64(__m512i __a, __m512i __b, __mmask8 *__m0, __mmask8 *__m1) {
+  __builtin_ia32_vp2intersect_q_512((__v8di)__a, (__v8di)__b, __m0, __m1);
+}
+
+#undef __DEFAULT_FN_ATTRS
+
+#endif

lib/Headers/immintrin.h

 #if !defined(_MSC_VER) || __has_feature(modules) || defined(__PTWRITE__)
 #include <ptwriteintrin.h>
 #endif
 
 #if !defined(_MSC_VER) || __has_feature(modules) || defined(__INVPCID__)
 #include <invpcidintrin.h>
 #endif
 
+#if !defined(_MSC_VER) || __has_feature(modules) || \
+  defined(__AVX512VP2INTERSECT__)
+#include <avx512vp2intersectintrin.h>
+#endif
+
+#if !defined(_MSC_VER) || __has_feature(modules) || \
+  (defined(__AVX512VL__) && defined(__AVX512VP2INTERSECT__))
+#include <avx512vlvp2intersectintrin.h>
+#endif
+
 #if defined(_MSC_VER) && __has_extension(gnu_asm)
 /* Define the default attributes for these intrinsics */
 #define __DEFAULT_FN_ATTRS __attribute__((__always_inline__, __nodebug__))
 #ifdef __cplusplus
 extern "C" {
 #endif
 /*----------------------------------------------------------------------------*\
 |* Interlocked Exchange HLE

test/CodeGen/attr-target-x86.c

 // CHECK: bar{{.*}} #0
 // CHECK: qux{{.*}} #1
 // CHECK: qax{{.*}} #5
 // CHECK: qq{{.*}} #6
 // CHECK: lake{{.*}} #7
 // CHECK: use_before_def{{.*}} #7
 // CHECK: #0 = {{.*}}"target-cpu"="i686" "target-features"="+cx8,+x87"
 // CHECK: #1 = {{.*}}"target-cpu"="ivybridge" "target-features"="+avx,+cx16,+cx8,+f16c,+fsgsbase,+fxsr,+mmx,+pclmul,+popcnt,+rdrnd,+sahf,+sse,+sse2,+sse3,+sse4.1,+sse4.2,+ssse3,+x87,+xsave,+xsaveopt"
-// CHECK: #2 = {{.*}}"target-cpu"="i686" "target-features"="+cx8,+x87,-aes,-avx,-avx2,-avx512bf16,-avx512bitalg,-avx512bw,-avx512cd,-avx512dq,-avx512er,-avx512f,-avx512ifma,-avx512pf,-avx512vbmi,-avx512vbmi2,-avx512vl,-avx512vnni,-avx512vpopcntdq,-f16c,-fma,-fma4,-gfni,-pclmul,-sha,-sse2,-sse3,-sse4.1,-sse4.2,-sse4a,-ssse3,-vaes,-vpclmulqdq,-xop,-xsave,-xsaveopt"
+// CHECK: #2 = {{.*}}"target-cpu"="i686" "target-features"="+cx8,+x87,-aes,-avx,-avx2,-avx512bf16,-avx512bitalg,-avx512bw,-avx512cd,-avx512dq,-avx512er,-avx512f,-avx512ifma,-avx512pf,-avx512vbmi,-avx512vbmi2,-avx512vl,-avx512vnni,-avx512vp2intersect,-avx512vpopcntdq,-f16c,-fma,-fma4,-gfni,-pclmul,-sha,-sse2,-sse3,-sse4.1,-sse4.2,-sse4a,-ssse3,-vaes,-vpclmulqdq,-xop,-xsave,-xsaveopt"
 // CHECK: #3 = {{.*}}"target-cpu"="i686" "target-features"="+cx8,+mmx,+popcnt,+sse,+sse2,+sse3,+sse4.1,+sse4.2,+ssse3,+x87"
-// CHECK: #4 = {{.*}}"target-cpu"="i686" "target-features"="+cx8,+x87,-avx,-avx2,-avx512bf16,-avx512bitalg,-avx512bw,-avx512cd,-avx512dq,-avx512er,-avx512f,-avx512ifma,-avx512pf,-avx512vbmi,-avx512vbmi2,-avx512vl,-avx512vnni,-avx512vpopcntdq,-f16c,-fma,-fma4,-sse4.1,-sse4.2,-vaes,-vpclmulqdq,-xop,-xsave,-xsaveopt"
+// CHECK: #4 = {{.*}}"target-cpu"="i686" "target-features"="+cx8,+x87,-avx,-avx2,-avx512bf16,-avx512bitalg,-avx512bw,-avx512cd,-avx512dq,-avx512er,-avx512f,-avx512ifma,-avx512pf,-avx512vbmi,-avx512vbmi2,-avx512vl,-avx512vnni,-avx512vp2intersect,-avx512vpopcntdq,-f16c,-fma,-fma4,-sse4.1,-sse4.2,-vaes,-vpclmulqdq,-xop,-xsave,-xsaveopt"
 // CHECK: #5 = {{.*}}"target-cpu"="ivybridge" "target-features"="+avx,+cx16,+cx8,+f16c,+fsgsbase,+fxsr,+mmx,+pclmul,+popcnt,+rdrnd,+sahf,+sse,+sse2,+sse3,+sse4.1,+sse4.2,+ssse3,+x87,+xsave,+xsaveopt,-aes,-vaes"
 // CHECK: #6 = {{.*}}"target-cpu"="i686" "target-features"="+cx8,+x87,-3dnow,-3dnowa,-mmx"
 // CHECK: #7 = {{.*}}"target-cpu"="lakemont" "target-features"="+cx8,+mmx"

test/CodeGen/intel-avx512vlvp2intersect.c

+// RUN: %clang_cc1 %s -ffreestanding -triple=x86_64-unknown-unknown -target-feature +avx512vp2intersect -target-feature +avx512vl -emit-llvm -o - -Wall -Werror | FileCheck %s
+// RUN: %clang_cc1 %s -ffreestanding -triple=i386-unknown-unknown -target-feature +avx512vp2intersect -target-feature +avx512vl -emit-llvm -o - -Wall -Werror | FileCheck %s
+
+#include <immintrin.h>
+
+void test_mm256_2intersect_epi32(__m256i a, __m256i b, __mmask8 *m0, __mmask8 *m1) {
+// CHECK-LABEL: test_mm256_2intersect_epi32
+// CHECK: call { <8 x i1>, <8 x i1> } @llvm.x86.avx512.vp2intersect.d.256(<8 x i32> %{{.*}}, <8 x i32> %{{.*}})
+// CHECK: extractvalue { <8 x i1>, <8 x i1> } %{{.*}}, 0
+// CHECK: extractvalue { <8 x i1>, <8 x i1> } %{{.*}}, 1
+  _mm256_2intersect_epi32(a, b, m0, m1);
+}
+
+void test_mm256_2intersect_epi64(__m256i a, __m256i b, __mmask8 *m0, __mmask8 *m1) {
+// CHECK-LABEL: test_mm256_2intersect_epi64
+// CHECK: call { <4 x i1>, <4 x i1> } @llvm.x86.avx512.vp2intersect.q.256(<4 x i64> %{{.*}}, <4 x i64> %{{.*}})
+// CHECK: extractvalue { <4 x i1>, <4 x i1> } %{{.*}}, 0
+// CHECK: extractvalue { <4 x i1>, <4 x i1> } %{{.*}}, 1
+  _mm256_2intersect_epi64(a, b, m0, m1);
+}
+
+void test_mm_2intersect_epi32(__m128i a, __m128i b, __mmask8 *m0, __mmask8 *m1) {
+// CHECK-LABEL: test_mm_2intersect_epi32
+// CHECK: call { <4 x i1>, <4 x i1> } @llvm.x86.avx512.vp2intersect.d.128(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
+// CHECK: extractvalue { <4 x i1>, <4 x i1> } %{{.*}}, 0
+// CHECK: extractvalue { <4 x i1>, <4 x i1> } %{{.*}}, 1
+  _mm_2intersect_epi32(a, b, m0, m1);
+}
+
+void test_mm_2intersect_epi64(__m128i a, __m128i b, __mmask8 *m0, __mmask8 *m1) {
+// CHECK-LABEL: test_mm_2intersect_epi64
+// CHECK: call { <2 x i1>, <2 x i1> } @llvm.x86.avx512.vp2intersect.q.128(<2 x i64> %{{.*}}, <2 x i64> %{{.*}})
+// CHECK: extractvalue { <2 x i1>, <2 x i1> } %{{.*}}, 0
+// CHECK: extractvalue { <2 x i1>, <2 x i1> } %{{.*}}, 1
+  _mm_2intersect_epi64(a, b, m0, m1);
+}

test/CodeGen/intel-avx512vp2intersect.c

+// RUN: %clang_cc1 %s -ffreestanding -triple=x86_64-unknown-unknown -target-feature +avx512vp2intersect -emit-llvm -o - -Wall -Werror | FileCheck %s
+// RUN: %clang_cc1 %s -ffreestanding -triple=i386-unknown-unknown -target-feature +avx512vp2intersect -emit-llvm -o - -Wall -Werror | FileCheck %s
+
+#include <immintrin.h>
+
+void test_mm512_2intersect_epi32(__m512i a, __m512i b, __mmask16 *m0, __mmask16 *m1) {
+// CHECK-LABEL: test_mm512_2intersect_epi32
+// CHECK: call { <16 x i1>, <16 x i1> } @llvm.x86.avx512.vp2intersect.d.512(<16 x i32> %{{.*}}, <16 x i32> %{{.*}})
+// CHECK: extractvalue { <16 x i1>, <16 x i1> } %{{.*}}, 0
+// CHECK: extractvalue { <16 x i1>, <16 x i1> } %{{.*}}, 1
+  _mm512_2intersect_epi32(a, b, m0, m1);
+}
+
+void test_mm512_2intersect_epi64(__m512i a, __m512i b, __mmask8 *m0, __mmask8 *m1) {
+// CHECK-LABEL: test_mm512_2intersect_epi64
+// CHECK: call { <8 x i1>, <8 x i1> } @llvm.x86.avx512.vp2intersect.q.512(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
+// CHECK: extractvalue { <8 x i1>, <8 x i1> } %{{.*}}, 0
+// CHECK: extractvalue { <8 x i1>, <8 x i1> } %{{.*}}, 1
+  _mm512_2intersect_epi64(a, b, m0, m1);
+}

test/Driver/x86-target-features.c

 // VNNI: "-target-feature" "+avx512vnni"
 // NO-VNNI: "-target-feature" "-avx512vnni"
 
 // RUN: %clang -target i386-unknown-linux-gnu -march=i386 -mavx512vbmi2 %s -### -o %t.o 2>&1 | FileCheck -check-prefix=VBMI2 %s
 // RUN: %clang -target i386-unknown-linux-gnu -march=i386 -mno-avx512vbmi2 %s -### -o %t.o 2>&1 | FileCheck -check-prefix=NO-VBMI2 %s
 // VBMI2: "-target-feature" "+avx512vbmi2"
 // NO-VBMI2: "-target-feature" "-avx512vbmi2"
 
+// RUN: %clang -target i386-linux-gnu -mavx512vp2intersect %s -### -o %t.o 2>&1 | FileCheck -check-prefix=VP2INTERSECT %s
+// RUN: %clang -target i386-linux-gnu -mno-avx512vp2intersect %s -### -o %t.o 2>&1 | FileCheck -check-prefix=NO-VP2INTERSECT %s
+// VP2INTERSECT: "-target-feature" "+avx512vp2intersect"
+// NO-VP2INTERSECT: "-target-feature" "-avx512vp2intersect"
+
 // RUN: %clang -target i386-unknown-linux-gnu -march=i386 -mrdpid %s -### -o %t.o 2>&1 | FileCheck -check-prefix=RDPID %s
 // RUN: %clang -target i386-unknown-linux-gnu -march=i386 -mno-rdpid %s -### -o %t.o 2>&1 | FileCheck -check-prefix=NO-RDPID %s
 // RDPID: "-target-feature" "+rdpid"
 // NO-RDPID: "-target-feature" "-rdpid"
 
 // RUN: %clang -target i386-linux-gnu -mretpoline %s -### -o %t.o 2>&1 | FileCheck -check-prefix=RETPOLINE %s
 // RUN: %clang -target i386-linux-gnu -mno-retpoline %s -### -o %t.o 2>&1 | FileCheck -check-prefix=NO-RETPOLINE %s
 // RETPOLINE: "-target-feature" "+retpoline-indirect-calls" "-target-feature" "+retpoline-indirect-branches"

test/Preprocessor/x86_target_features.c

 // RUN: %clang -target i386-unknown-unknown -march=atom -mavx512bf16 -mno-avx512bw -x c -E -dM -o - %s | FileCheck -match-full-lines --check-prefix=AVX512BF16_NOAVX512BW %s
 
 // AVX512BF16_NOAVX512BW-NOT: #define __AVX512BF16__ 1
 
 // RUN: %clang -target i386-unknown-unknown -march=atom -mavx512bf16 -mno-avx512vl -x c -E -dM -o - %s | FileCheck -match-full-lines --check-prefix=AVX512BF16_NOAVX512VL %s
 
 // AVX512BF16_NOAVX512VL: #define __AVX512BF16__ 1
 
+// RUN: %clang -target i386-unknown-linux-gnu -march=i386 -mavx512vp2intersect -x c -E -dM -o - %s | FileCheck  -check-prefix=VP2INTERSECT %s
+
+// VP2INTERSECT: #define __AVX512F__ 1
+// VP2INTERSECT: #define __AVX512VP2INTERSECT__ 1
+
+// RUN: %clang -target i386-unknown-linux-gnu -march=i386 -mno-avx512vp2intersect -x c -E -dM -o - %s | FileCheck  -check-prefix=NOVP2INTERSECT %s
+// RUN: %clang -target i386-unknown-linux-gnu -march=i386 -mavx512vp2intersect -mno-avx512f -x c -E -dM -o - %s | FileCheck  -check-prefix=NOVP2INTERSECT %s
+
+// NOVP2INTERSECT-NOT: #define __AVX512VP2INTERSECT__ 1
+