[X86] Convert integer _mm_reduce_* intrinsics to emit llvm.reduction intrinsics (PR47506)

[RKSimon]

Summary

Placeholder patch for when the reduction intrinsics drop their experimental status, emitting the equivalent reduction intrinsic in IR instead of expanding to shuffle+arithmetic sequences.

The fadd/fmul reductions might be trickier as they assume a similar bisection reduction while the generic intrinsics assume a sequential reduction (intel docs are ambiguous on the correct approach) - I'm not sure if we want to always tag them with reassoc? Anyway, that issue can wait until a separate fp patch along with the fmin/fmax reductions.

[craig.topper]

Longer term should we have a variadic version of this like builtin_shufflevector or builtin_convertvector that can handle any reduction?

[RKSimon]

In D87604#2272211, @craig.topper wrote:

Longer term should we have a variadic version of this like builtin_shufflevector or builtin_convertvector that can handle any reduction?

Yes, I was looking at what would be necessary for that as well - I think that is what is proposed in PR36691.

[RKSimon]

WIP until reductions are no longer experimental

[RKSimon]

rebased now that the reduction intrinsics are no longer experiemental

[RKSimon]

Ping?

[craig.topper]

LGTM

clang/include/clang/Basic/BuiltinsX86.def

 TARGET_BUILTIN(__builtin_ia32_selectps_256, "V8fUcV8fV8f", "ncV:256:", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_selectps_512, "V16fUsV16fV16f", "ncV:512:", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_selectpd_128, "V2dUcV2dV2d", "ncV:128:", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_selectpd_256, "V4dUcV4dV4d", "ncV:256:", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_selectpd_512, "V8dUcV8dV8d", "ncV:512:", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_selectss_128, "V4fUcV4fV4f", "ncV:128:", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_selectsd_128, "V2dUcV2dV2d", "ncV:128:", "avx512f")
 
+// generic reduction intrinsics
+TARGET_BUILTIN(__builtin_ia32_reduce_add_d512, "iV16i", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_add_q512, "OiV8Oi", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_and_d512, "iV16i", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_and_q512, "OiV8Oi", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_mul_d512, "iV16i", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_mul_q512, "OiV8Oi", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_or_d512, "iV16i", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_or_q512, "OiV8Oi", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_smax_d512, "iV16i", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_smax_q512, "OiV8Oi", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_smin_d512, "iV16i", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_smin_q512, "OiV8Oi", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_umax_d512, "iV16i", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_umax_q512, "OiV8Oi", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_umin_d512, "iV16i", "ncV:512:", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_reduce_umin_q512, "OiV8Oi", "ncV:512:", "avx512f")
+
 // MONITORX/MWAITX
 TARGET_BUILTIN(__builtin_ia32_monitorx, "vvC*UiUi", "n", "mwaitx")
 TARGET_BUILTIN(__builtin_ia32_mwaitx, "vUiUiUi", "n", "mwaitx")
 
 // WAITPKG
 TARGET_BUILTIN(__builtin_ia32_umonitor, "vvC*", "n", "waitpkg")
 TARGET_BUILTIN(__builtin_ia32_umwait, "UcUiUiUi", "n", "waitpkg")
 TARGET_BUILTIN(__builtin_ia32_tpause, "UcUiUiUi", "n", "waitpkg")

clang/lib/CodeGen/CGBuiltin.cpp

   case X86::BI__builtin_ia32_vpshrdvq256:
   case X86::BI__builtin_ia32_vpshrdvq512:
   case X86::BI__builtin_ia32_vpshrdvw128:
   case X86::BI__builtin_ia32_vpshrdvw256:
   case X86::BI__builtin_ia32_vpshrdvw512:
     // Ops 0 and 1 are swapped.
     return EmitX86FunnelShift(*this, Ops[1], Ops[0], Ops[2], true);
 
+  // Reductions
+  case X86::BI__builtin_ia32_reduce_add_d512:
+  case X86::BI__builtin_ia32_reduce_add_q512: {
+    Function *F =
+        CGM.getIntrinsic(Intrinsic::vector_reduce_add, Ops[0]->getType());
+    return Builder.CreateCall(F, {Ops[0]});
+  }
+  case X86::BI__builtin_ia32_reduce_and_d512:
+  case X86::BI__builtin_ia32_reduce_and_q512: {
+    Function *F =
+        CGM.getIntrinsic(Intrinsic::vector_reduce_and, Ops[0]->getType());
+    return Builder.CreateCall(F, {Ops[0]});
+  }
+  case X86::BI__builtin_ia32_reduce_mul_d512:
+  case X86::BI__builtin_ia32_reduce_mul_q512: {
+    Function *F =
+        CGM.getIntrinsic(Intrinsic::vector_reduce_mul, Ops[0]->getType());
+    return Builder.CreateCall(F, {Ops[0]});
+  }
+  case X86::BI__builtin_ia32_reduce_or_d512:
+  case X86::BI__builtin_ia32_reduce_or_q512: {
+    Function *F =
+        CGM.getIntrinsic(Intrinsic::vector_reduce_or, Ops[0]->getType());
+    return Builder.CreateCall(F, {Ops[0]});
+  }
+  case X86::BI__builtin_ia32_reduce_smax_d512:
+  case X86::BI__builtin_ia32_reduce_smax_q512: {
+    Function *F =
+        CGM.getIntrinsic(Intrinsic::vector_reduce_smax, Ops[0]->getType());
+    return Builder.CreateCall(F, {Ops[0]});
+  }
+  case X86::BI__builtin_ia32_reduce_smin_d512:
+  case X86::BI__builtin_ia32_reduce_smin_q512: {
+    Function *F =
+        CGM.getIntrinsic(Intrinsic::vector_reduce_smin, Ops[0]->getType());
+    return Builder.CreateCall(F, {Ops[0]});
+  }
+  case X86::BI__builtin_ia32_reduce_umax_d512:
+  case X86::BI__builtin_ia32_reduce_umax_q512: {
+    Function *F =
+        CGM.getIntrinsic(Intrinsic::vector_reduce_umax, Ops[0]->getType());
+    return Builder.CreateCall(F, {Ops[0]});
+  }
+  case X86::BI__builtin_ia32_reduce_umin_d512:
+  case X86::BI__builtin_ia32_reduce_umin_q512: {
+    Function *F =
+        CGM.getIntrinsic(Intrinsic::vector_reduce_umin, Ops[0]->getType());
+    return Builder.CreateCall(F, {Ops[0]});
+  }
+
   // 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");
   }

clang/lib/Headers/avx512fintrin.h

  * computations. In vector-reduction arithmetic, the evaluation off is
  * independent of the order of the input elements of V.
 
  * Used bisection method. At each step, we partition the vector with previous
  * step in half, and the operation is performed on its two halves.
  * This takes log2(n) steps where n is the number of elements in the vector.
  */
 
-#define _mm512_mask_reduce_operator(op) \
-  __v4du __t1 = (__v4du)_mm512_extracti64x4_epi64(__W, 0); \
-  __v4du __t2 = (__v4du)_mm512_extracti64x4_epi64(__W, 1); \
-  __m256i __t3 = (__m256i)(__t1 op __t2); \
-  __v2du __t4 = (__v2du)_mm256_extracti128_si256(__t3, 0); \
-  __v2du __t5 = (__v2du)_mm256_extracti128_si256(__t3, 1); \
-  __v2du __t6 = __t4 op __t5; \
-  __v2du __t7 = __builtin_shufflevector(__t6, __t6, 1, 0); \
-  __v2du __t8 = __t6 op __t7; \
-  return __t8[0]
-
 static __inline__ long long __DEFAULT_FN_ATTRS512 _mm512_reduce_add_epi64(__m512i __W) {
-  _mm512_mask_reduce_operator(+);
+  return __builtin_ia32_reduce_add_q512(__W);
 }
 
 static __inline__ long long __DEFAULT_FN_ATTRS512 _mm512_reduce_mul_epi64(__m512i __W) {
-  _mm512_mask_reduce_operator(*);
+  return __builtin_ia32_reduce_mul_q512(__W);
 }
 
 static __inline__ long long __DEFAULT_FN_ATTRS512 _mm512_reduce_and_epi64(__m512i __W) {
-  _mm512_mask_reduce_operator(&);
+  return __builtin_ia32_reduce_and_q512(__W);
 }
 
 static __inline__ long long __DEFAULT_FN_ATTRS512 _mm512_reduce_or_epi64(__m512i __W) {
-  _mm512_mask_reduce_operator(|);
+  return __builtin_ia32_reduce_or_q512(__W);
 }
 
 static __inline__ long long __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_add_epi64(__mmask8 __M, __m512i __W) {
   __W = _mm512_maskz_mov_epi64(__M, __W);
-  _mm512_mask_reduce_operator(+);
+  return __builtin_ia32_reduce_add_q512(__W);
 }
 
 static __inline__ long long __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_mul_epi64(__mmask8 __M, __m512i __W) {
   __W = _mm512_mask_mov_epi64(_mm512_set1_epi64(1), __M, __W);
-  _mm512_mask_reduce_operator(*);
+  return __builtin_ia32_reduce_mul_q512(__W);
 }
 
 static __inline__ long long __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_and_epi64(__mmask8 __M, __m512i __W) {
   __W = _mm512_mask_mov_epi64(_mm512_set1_epi64(~0ULL), __M, __W);
-  _mm512_mask_reduce_operator(&);
+  return __builtin_ia32_reduce_and_q512(__W);
 }
 
 static __inline__ long long __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_or_epi64(__mmask8 __M, __m512i __W) {
   __W = _mm512_maskz_mov_epi64(__M, __W);
-  _mm512_mask_reduce_operator(|);
+  return __builtin_ia32_reduce_or_q512(__W);
 }
-#undef _mm512_mask_reduce_operator
 
 #define _mm512_mask_reduce_operator(op) \
   __m256d __t1 = _mm512_extractf64x4_pd(__W, 0); \
   __m256d __t2 = _mm512_extractf64x4_pd(__W, 1); \
   __m256d __t3 = __t1 op __t2; \
   __m128d __t4 = _mm256_extractf128_pd(__t3, 0); \
   __m128d __t5 = _mm256_extractf128_pd(__t3, 1); \
   __m128d __t6 = __t4 op __t5; \

 
 static __inline__ double __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_mul_pd(__mmask8 __M, __m512d __W) {
   __W = _mm512_mask_mov_pd(_mm512_set1_pd(1.0), __M, __W);
   _mm512_mask_reduce_operator(*);
 }
 #undef _mm512_mask_reduce_operator
 
-#define _mm512_mask_reduce_operator(op) \
-  __v8su __t1 = (__v8su)_mm512_extracti64x4_epi64(__W, 0); \
-  __v8su __t2 = (__v8su)_mm512_extracti64x4_epi64(__W, 1); \
-  __m256i __t3 = (__m256i)(__t1 op __t2); \
-  __v4su __t4 = (__v4su)_mm256_extracti128_si256(__t3, 0); \
-  __v4su __t5 = (__v4su)_mm256_extracti128_si256(__t3, 1); \
-  __v4su __t6 = __t4 op __t5; \
-  __v4su __t7 = __builtin_shufflevector(__t6, __t6, 2, 3, 0, 1); \
-  __v4su __t8 = __t6 op __t7; \
-  __v4su __t9 = __builtin_shufflevector(__t8, __t8, 1, 0, 3, 2); \
-  __v4su __t10 = __t8 op __t9; \
-  return __t10[0]
-
 static __inline__ int __DEFAULT_FN_ATTRS512
 _mm512_reduce_add_epi32(__m512i __W) {
-  _mm512_mask_reduce_operator(+);
+  return __builtin_ia32_reduce_add_d512((__v16si)__W);
 }
 
 static __inline__ int __DEFAULT_FN_ATTRS512
 _mm512_reduce_mul_epi32(__m512i __W) {
-  _mm512_mask_reduce_operator(*);
+  return __builtin_ia32_reduce_mul_d512((__v16si)__W);
 }
 
 static __inline__ int __DEFAULT_FN_ATTRS512
 _mm512_reduce_and_epi32(__m512i __W) {
-  _mm512_mask_reduce_operator(&);
+  return __builtin_ia32_reduce_and_d512((__v16si)__W);
 }
 
 static __inline__ int __DEFAULT_FN_ATTRS512
 _mm512_reduce_or_epi32(__m512i __W) {
-  _mm512_mask_reduce_operator(|);
+  return __builtin_ia32_reduce_or_d512((__v16si)__W);
 }
 
 static __inline__ int __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_add_epi32( __mmask16 __M, __m512i __W) {
   __W = _mm512_maskz_mov_epi32(__M, __W);
-  _mm512_mask_reduce_operator(+);
+  return __builtin_ia32_reduce_add_d512((__v16si)__W);
 }
 
 static __inline__ int __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_mul_epi32( __mmask16 __M, __m512i __W) {
   __W = _mm512_mask_mov_epi32(_mm512_set1_epi32(1), __M, __W);
-  _mm512_mask_reduce_operator(*);
+  return __builtin_ia32_reduce_mul_d512((__v16si)__W);
 }
 
 static __inline__ int __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_and_epi32( __mmask16 __M, __m512i __W) {
   __W = _mm512_mask_mov_epi32(_mm512_set1_epi32(~0U), __M, __W);
-  _mm512_mask_reduce_operator(&);
+  return __builtin_ia32_reduce_and_d512((__v16si)__W);
 }
 
 static __inline__ int __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_or_epi32(__mmask16 __M, __m512i __W) {
   __W = _mm512_maskz_mov_epi32(__M, __W);
-  _mm512_mask_reduce_operator(|);
+  return __builtin_ia32_reduce_or_d512((__v16si)__W);
 }
-#undef _mm512_mask_reduce_operator
 
 #define _mm512_mask_reduce_operator(op) \
   __m256 __t1 = (__m256)_mm512_extractf64x4_pd((__m512d)__W, 0); \
   __m256 __t2 = (__m256)_mm512_extractf64x4_pd((__m512d)__W, 1); \
   __m256 __t3 = __t1 op __t2; \
   __m128 __t4 = _mm256_extractf128_ps(__t3, 0); \
   __m128 __t5 = _mm256_extractf128_ps(__t3, 1); \
   __m128 __t6 = __t4 op __t5; \

 
 static __inline__ float __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_mul_ps(__mmask16 __M, __m512 __W) {
   __W = _mm512_mask_mov_ps(_mm512_set1_ps(1.0f), __M, __W);
   _mm512_mask_reduce_operator(*);
 }
 #undef _mm512_mask_reduce_operator
 
-#define _mm512_mask_reduce_operator(op) \
-  __m512i __t1 = (__m512i)__builtin_shufflevector((__v8di)__V, (__v8di)__V, 4, 5, 6, 7, 0, 1, 2, 3); \
-  __m512i __t2 = _mm512_##op(__V, __t1); \
-  __m512i __t3 = (__m512i)__builtin_shufflevector((__v8di)__t2, (__v8di)__t2, 2, 3, 0, 1, 6, 7, 4, 5); \
-  __m512i __t4 = _mm512_##op(__t2, __t3); \
-  __m512i __t5 = (__m512i)__builtin_shufflevector((__v8di)__t4, (__v8di)__t4, 1, 0, 3, 2, 5, 4, 7, 6); \
-  __v8di __t6 = (__v8di)_mm512_##op(__t4, __t5); \
-  return __t6[0]
-
 static __inline__ long long __DEFAULT_FN_ATTRS512
 _mm512_reduce_max_epi64(__m512i __V) {
-  _mm512_mask_reduce_operator(max_epi64);
+  return __builtin_ia32_reduce_smax_q512(__V);
 }
 
 static __inline__ unsigned long long __DEFAULT_FN_ATTRS512
 _mm512_reduce_max_epu64(__m512i __V) {
-  _mm512_mask_reduce_operator(max_epu64);
+  return __builtin_ia32_reduce_umax_q512(__V);
 }
 
 static __inline__ long long __DEFAULT_FN_ATTRS512
 _mm512_reduce_min_epi64(__m512i __V) {
-  _mm512_mask_reduce_operator(min_epi64);
+  return __builtin_ia32_reduce_smin_q512(__V);
 }
 
 static __inline__ unsigned long long __DEFAULT_FN_ATTRS512
 _mm512_reduce_min_epu64(__m512i __V) {
-  _mm512_mask_reduce_operator(min_epu64);
+  return __builtin_ia32_reduce_umin_q512(__V);
 }
 
 static __inline__ long long __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_max_epi64(__mmask8 __M, __m512i __V) {
   __V = _mm512_mask_mov_epi64(_mm512_set1_epi64(-__LONG_LONG_MAX__ - 1LL), __M, __V);
-  _mm512_mask_reduce_operator(max_epi64);
+  return __builtin_ia32_reduce_smax_q512(__V);
 }
 
 static __inline__ unsigned long long __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_max_epu64(__mmask8 __M, __m512i __V) {
   __V = _mm512_maskz_mov_epi64(__M, __V);
-  _mm512_mask_reduce_operator(max_epu64);
+  return __builtin_ia32_reduce_umax_q512(__V);
 }
 
 static __inline__ long long __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_min_epi64(__mmask8 __M, __m512i __V) {
   __V = _mm512_mask_mov_epi64(_mm512_set1_epi64(__LONG_LONG_MAX__), __M, __V);
-  _mm512_mask_reduce_operator(min_epi64);
+  return __builtin_ia32_reduce_smin_q512(__V);
 }
 
 static __inline__ unsigned long long __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_min_epu64(__mmask8 __M, __m512i __V) {
   __V = _mm512_mask_mov_epi64(_mm512_set1_epi64(~0ULL), __M, __V);
-  _mm512_mask_reduce_operator(min_epu64);
+  return __builtin_ia32_reduce_umin_q512(__V);
 }
-#undef _mm512_mask_reduce_operator
-
-#define _mm512_mask_reduce_operator(op) \
-  __m256i __t1 = _mm512_extracti64x4_epi64(__V, 0); \
-  __m256i __t2 = _mm512_extracti64x4_epi64(__V, 1); \
-  __m256i __t3 = _mm256_##op(__t1, __t2); \
-  __m128i __t4 = _mm256_extracti128_si256(__t3, 0); \
-  __m128i __t5 = _mm256_extracti128_si256(__t3, 1); \
-  __m128i __t6 = _mm_##op(__t4, __t5); \
-  __m128i __t7 = (__m128i)__builtin_shufflevector((__v4si)__t6, (__v4si)__t6, 2, 3, 0, 1); \
-  __m128i __t8 = _mm_##op(__t6, __t7); \
-  __m128i __t9 = (__m128i)__builtin_shufflevector((__v4si)__t8, (__v4si)__t8, 1, 0, 3, 2); \
-  __v4si __t10 = (__v4si)_mm_##op(__t8, __t9); \
-  return __t10[0]
-
 static __inline__ int __DEFAULT_FN_ATTRS512
 _mm512_reduce_max_epi32(__m512i __V) {
-  _mm512_mask_reduce_operator(max_epi32);
+  return __builtin_ia32_reduce_smax_d512((__v16si)__V);
 }
 
 static __inline__ unsigned int __DEFAULT_FN_ATTRS512
 _mm512_reduce_max_epu32(__m512i __V) {
-  _mm512_mask_reduce_operator(max_epu32);
+  return __builtin_ia32_reduce_umax_d512((__v16si)__V);
 }
 
 static __inline__ int __DEFAULT_FN_ATTRS512
 _mm512_reduce_min_epi32(__m512i __V) {
-  _mm512_mask_reduce_operator(min_epi32);
+  return __builtin_ia32_reduce_smin_d512((__v16si)__V);
 }
 
 static __inline__ unsigned int __DEFAULT_FN_ATTRS512
 _mm512_reduce_min_epu32(__m512i __V) {
-  _mm512_mask_reduce_operator(min_epu32);
+  return __builtin_ia32_reduce_umin_d512((__v16si)__V);
 }
 
 static __inline__ int __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_max_epi32(__mmask16 __M, __m512i __V) {
   __V = _mm512_mask_mov_epi32(_mm512_set1_epi32(-__INT_MAX__ - 1), __M, __V);
-  _mm512_mask_reduce_operator(max_epi32);
+  return __builtin_ia32_reduce_smax_d512((__v16si)__V);
 }
 
 static __inline__ unsigned int __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_max_epu32(__mmask16 __M, __m512i __V) {
   __V = _mm512_maskz_mov_epi32(__M, __V);
-  _mm512_mask_reduce_operator(max_epu32);
+  return __builtin_ia32_reduce_umax_d512((__v16si)__V);
 }
 
 static __inline__ int __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_min_epi32(__mmask16 __M, __m512i __V) {
   __V = _mm512_mask_mov_epi32(_mm512_set1_epi32(__INT_MAX__), __M, __V);
-  _mm512_mask_reduce_operator(min_epi32);
+  return __builtin_ia32_reduce_smin_d512((__v16si)__V);
 }
 
 static __inline__ unsigned int __DEFAULT_FN_ATTRS512
 _mm512_mask_reduce_min_epu32(__mmask16 __M, __m512i __V) {
   __V = _mm512_mask_mov_epi32(_mm512_set1_epi32(~0U), __M, __V);
-  _mm512_mask_reduce_operator(min_epu32);
+  return __builtin_ia32_reduce_umin_d512((__v16si)__V);
 }
-#undef _mm512_mask_reduce_operator
 
 #define _mm512_mask_reduce_operator(op) \
   __m256d __t1 = _mm512_extractf64x4_pd(__V, 0); \
   __m256d __t2 = _mm512_extractf64x4_pd(__V, 1); \
   __m256d __t3 = _mm256_##op(__t1, __t2); \
   __m128d __t4 = _mm256_extractf128_pd(__t3, 0); \
   __m128d __t5 = _mm256_extractf128_pd(__t3, 1); \
   __m128d __t6 = _mm_##op(__t4, __t5); \

clang/test/CodeGen/X86/avx512-reduceIntrin.c

 // RUN: %clang_cc1 -ffreestanding %s -O0 -triple=x86_64-apple-darwin -target-cpu skylake-avx512 -emit-llvm -o - -Wall -Werror | FileCheck %s
 
 #include <immintrin.h>
 
 long long test_mm512_reduce_add_epi64(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_add_epi64(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    add <4 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    add <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <2 x i64> %{{.*}}, <2 x i64> %{{.*}}, <2 x i32> <i32 1, i32 0>
-// CHECK:    add <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <2 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %{{.*}})
   return _mm512_reduce_add_epi64(__W);
 }
 
 long long test_mm512_reduce_mul_epi64(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_mul_epi64(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    mul <4 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    mul <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <2 x i64> %{{.*}}, <2 x i64> %{{.*}}, <2 x i32> <i32 1, i32 0>
-// CHECK:    mul <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <2 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.mul.v8i64(<8 x i64> %{{.*}})
   return _mm512_reduce_mul_epi64(__W); 
 }
 
 long long test_mm512_reduce_or_epi64(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_or_epi64(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    or <4 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    or <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <2 x i64> %{{.*}}, <2 x i64> %{{.*}}, <2 x i32> <i32 1, i32 0>
-// CHECK:    or <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <2 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.or.v8i64(<8 x i64> %{{.*}})
   return _mm512_reduce_or_epi64(__W); 
 }
 
 long long test_mm512_reduce_and_epi64(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_and_epi64(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    and <4 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    and <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <2 x i64> %{{.*}}, <2 x i64> %{{.*}}, <2 x i32> <i32 1, i32 0>
-// CHECK:    and <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <2 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.and.v8i64(<8 x i64> %{{.*}})
   return _mm512_reduce_and_epi64(__W);
 }
 
 long long test_mm512_mask_reduce_add_epi64(__mmask8 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_add_epi64(
 // CHECK:    bitcast i8 %{{.*}} to <8 x i1>
 // CHECK:    select <8 x i1> %{{.*}}, <8 x i64> %{{.*}}, <8 x i64> %{{.*}}
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    add <4 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    add <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <2 x i64> %{{.*}}, <2 x i64> %{{.*}}, <2 x i32> <i32 1, i32 0>
-// CHECK:    add <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <2 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %{{.*}})
   return _mm512_mask_reduce_add_epi64(__M, __W); 
 }
 
 long long test_mm512_mask_reduce_mul_epi64(__mmask8 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_mul_epi64(
 // CHECK:    bitcast i8 %{{.*}} to <8 x i1>
 // CHECK:    select <8 x i1> %{{.*}}, <8 x i64> %{{.*}}, <8 x i64> %{{.*}}
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    mul <4 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    mul <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <2 x i64> %{{.*}}, <2 x i64> %{{.*}}, <2 x i32> <i32 1, i32 0>
-// CHECK:    mul <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <2 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.mul.v8i64(<8 x i64> %{{.*}})
   return _mm512_mask_reduce_mul_epi64(__M, __W); 
 }
 
 long long test_mm512_mask_reduce_and_epi64(__mmask8 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_and_epi64(
 // CHECK:    bitcast i8 %{{.*}} to <8 x i1>
 // CHECK:    select <8 x i1> %{{.*}}, <8 x i64> %{{.*}}, <8 x i64> %{{.*}}
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    and <4 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    and <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <2 x i64> %{{.*}}, <2 x i64> %{{.*}}, <2 x i32> <i32 1, i32 0>
-// CHECK:    and <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <2 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.and.v8i64(<8 x i64> %{{.*}})
   return _mm512_mask_reduce_and_epi64(__M, __W); 
 }
 
 long long test_mm512_mask_reduce_or_epi64(__mmask8 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_or_epi64(
 // CHECK:    bitcast i8 %{{.*}} to <8 x i1>
 // CHECK:    select <8 x i1> %{{.*}}, <8 x i64> %{{.*}}, <8 x i64> %{{.*}}
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    or <4 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    or <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <2 x i64> %{{.*}}, <2 x i64> %{{.*}}, <2 x i32> <i32 1, i32 0>
-// CHECK:    or <2 x i64> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <2 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.or.v8i64(<8 x i64> %{{.*}})
   return _mm512_mask_reduce_or_epi64(__M, __W); 
 }
 
 int test_mm512_reduce_add_epi32(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_add_epi32(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    add <8 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    add <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    add <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    add <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %{{.*}})
   return _mm512_reduce_add_epi32(__W); 
 }
 
 int test_mm512_reduce_mul_epi32(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_mul_epi32(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    mul <8 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    mul <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    mul <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    mul <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.mul.v16i32(<16 x i32> %{{.*}})
   return _mm512_reduce_mul_epi32(__W); 
 }
 
 int test_mm512_reduce_or_epi32(__m512i __W){
-// CHECK-LABEL: @test_mm512_reduce_or_epi32(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    or <8 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    or <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    or <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    or <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.or.v16i32(<16 x i32> %{{.*}})
   return _mm512_reduce_or_epi32(__W); 
 }
 
 int test_mm512_reduce_and_epi32(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_and_epi32(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    and <8 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    and <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    and <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    and <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.and.v16i32(<16 x i32> %{{.*}})
   return _mm512_reduce_and_epi32(__W); 
 }
 
 int test_mm512_mask_reduce_add_epi32(__mmask16 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_add_epi32(
 // CHECK:    bitcast i16 %{{.*}} to <16 x i1>
 // CHECK:    select <16 x i1> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> %{{.*}}
-// CHECK:    bitcast <16 x i32> %{{.*}} to <8 x i64>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    add <8 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    add <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    add <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    add <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %{{.*}})
   return _mm512_mask_reduce_add_epi32(__M, __W); 
 }
 
 int test_mm512_mask_reduce_mul_epi32(__mmask16 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_mul_epi32(
 // CHECK:    bitcast i16 %{{.*}} to <16 x i1>
 // CHECK:    select <16 x i1> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> %{{.*}}
-// CHECK:    bitcast <16 x i32> %{{.*}} to <8 x i64>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    mul <8 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    mul <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    mul <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    mul <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.mul.v16i32(<16 x i32> %{{.*}})
   return _mm512_mask_reduce_mul_epi32(__M, __W); 
 }
 
 int test_mm512_mask_reduce_and_epi32(__mmask16 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_and_epi32(
 // CHECK:    bitcast i16 %{{.*}} to <16 x i1>
 // CHECK:    select <16 x i1> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> %{{.*}}
-// CHECK:    bitcast <16 x i32> %{{.*}} to <8 x i64>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    and <8 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    and <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    and <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    and <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.and.v16i32(<16 x i32> %{{.*}})
   return _mm512_mask_reduce_and_epi32(__M, __W); 
 }
 
 int test_mm512_mask_reduce_or_epi32(__mmask16 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_or_epi32(
 // CHECK:    bitcast i16 %{{.*}} to <16 x i1>
 // CHECK:    select <16 x i1> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> %{{.*}}
-// CHECK:    bitcast <16 x i32> %{{.*}} to <8 x i64>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    or <8 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    or <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    or <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    or <4 x i32> %{{.*}}, %{{.*}}
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.or.v16i32(<16 x i32> %{{.*}})
   return _mm512_mask_reduce_or_epi32(__M, __W); 
 }
 
 double test_mm512_reduce_add_pd(__m512d __W){
 // CHECK-LABEL: @test_mm512_reduce_add_pd(
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
 // CHECK:    fadd <4 x double> %{{.*}}, %{{.*}}

clang/test/CodeGen/X86/avx512-reduceMinMaxIntrin.c

 // RUN: %clang_cc1 -fexperimental-new-pass-manager -ffreestanding %s -O0 -triple=x86_64-apple-darwin -target-cpu skylake-avx512 -emit-llvm -o - -Wall -Werror | FileCheck %s
 
 #include <immintrin.h>
 
 long long test_mm512_reduce_max_epi64(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_max_epi64(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
-// CHECK:    call <8 x i64> @llvm.smax.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 5>
-// CHECK:    call <8 x i64> @llvm.smax.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
-// CHECK:    call <8 x i64> @llvm.smax.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    extractelement <8 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> %{{.*}})
   return _mm512_reduce_max_epi64(__W);
 }
 
 unsigned long long test_mm512_reduce_max_epu64(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_max_epu64(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
-// CHECK:    call <8 x i64> @llvm.umax.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 5>
-// CHECK:    call <8 x i64> @llvm.umax.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
-// CHECK:    call <8 x i64> @llvm.umax.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    extractelement <8 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> %{{.*}})
   return _mm512_reduce_max_epu64(__W);
 }
 
 double test_mm512_reduce_max_pd(__m512d __W){
 // CHECK-LABEL: @test_mm512_reduce_max_pd(
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
 // CHECK:    call <4 x double> @llvm.x86.avx.max.pd.256(<4 x double> %{{.*}}, <4 x double> %{{.*}})
 // CHECK:    shufflevector <4 x double> %{{.*}}, <4 x double> undef, <2 x i32> <i32 0, i32 1>
 // CHECK:    shufflevector <4 x double> %{{.*}}, <4 x double> undef, <2 x i32> <i32 2, i32 3>
 // CHECK:    call <2 x double> @llvm.x86.sse2.max.pd(<2 x double> %{{.*}}, <2 x double> %{{.*}})
 // CHECK:    shufflevector <2 x double> %{{.*}}, <2 x double> %{{.*}}, <2 x i32> <i32 1, i32 0>
 // CHECK:    call <2 x double> @llvm.x86.sse2.max.pd(<2 x double> %{{.*}}, <2 x double> %{{.*}})
 // CHECK:    extractelement <2 x double> %{{.*}}, i32 0
   return _mm512_reduce_max_pd(__W); 
 }
 
 long long test_mm512_reduce_min_epi64(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_min_epi64(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
-// CHECK:    call <8 x i64> @llvm.smin.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 5>
-// CHECK:    call <8 x i64> @llvm.smin.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
-// CHECK:    call <8 x i64> @llvm.smin.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    extractelement <8 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> %{{.*}})
   return _mm512_reduce_min_epi64(__W);
 }
 
 unsigned long long test_mm512_reduce_min_epu64(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_min_epu64(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
-// CHECK:    call <8 x i64> @llvm.umin.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 5>
-// CHECK:    call <8 x i64> @llvm.umin.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
-// CHECK:    call <8 x i64> @llvm.umin.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    extractelement <8 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> %{{.*}})
   return _mm512_reduce_min_epu64(__W);
 }
 
 double test_mm512_reduce_min_pd(__m512d __W){
 // CHECK-LABEL: @test_mm512_reduce_min_pd(
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
 // CHECK:    call <4 x double> @llvm.x86.avx.min.pd.256(<4 x double> %{{.*}}, <4 x double> %{{.*}})
 // CHECK:    shufflevector <4 x double> %{{.*}}, <4 x double> undef, <2 x i32> <i32 0, i32 1>
 // CHECK:    shufflevector <4 x double> %{{.*}}, <4 x double> undef, <2 x i32> <i32 2, i32 3>
 // CHECK:    call <2 x double> @llvm.x86.sse2.min.pd(<2 x double> %{{.*}}, <2 x double> %{{.*}})
 // CHECK:    shufflevector <2 x double> %{{.*}}, <2 x double> %{{.*}}, <2 x i32> <i32 1, i32 0>
 // CHECK:    call <2 x double> @llvm.x86.sse2.min.pd(<2 x double> %{{.*}}, <2 x double> %{{.*}})
 // CHECK:    extractelement <2 x double> %{{.*}}, i32 0
   return _mm512_reduce_min_pd(__W); 
 }
 
 long long test_mm512_mask_reduce_max_epi64(__mmask8 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_max_epi64(
 // CHECK:    bitcast i8 %{{.*}} to <8 x i1>
 // CHECK:    select <8 x i1> %{{.*}}, <8 x i64> %{{.*}}, <8 x i64> %{{.*}}
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
-// CHECK:    call <8 x i64> @llvm.smax.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 5>
-// CHECK:    call <8 x i64> @llvm.smax.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
-// CHECK:    call <8 x i64> @llvm.smax.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    extractelement <8 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> %{{.*}})
   return _mm512_mask_reduce_max_epi64(__M, __W); 
 }
 
 unsigned long test_mm512_mask_reduce_max_epu64(__mmask8 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_max_epu64(
 // CHECK:    bitcast i8 %{{.*}} to <8 x i1>
 // CHECK:    select <8 x i1> %{{.*}}, <8 x i64> %{{.*}}, <8 x i64> %{{.*}}
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
-// CHECK:    call <8 x i64> @llvm.umax.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 5>
-// CHECK:    call <8 x i64> @llvm.umax.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
-// CHECK:    call <8 x i64> @llvm.umax.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    extractelement <8 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> %{{.*}})
   return _mm512_mask_reduce_max_epu64(__M, __W); 
 }
 
 double test_mm512_mask_reduce_max_pd(__mmask8 __M, __m512d __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_max_pd(
 // CHECK:    bitcast i8 %{{.*}} to <8 x i1>
 // CHECK:    select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
 // CHECK:    call <4 x double> @llvm.x86.avx.max.pd.256(<4 x double> %{{.*}}, <4 x double> %{{.*}})
 // CHECK:    shufflevector <4 x double> %{{.*}}, <4 x double> undef, <2 x i32> <i32 0, i32 1>
 // CHECK:    shufflevector <4 x double> %{{.*}}, <4 x double> undef, <2 x i32> <i32 2, i32 3>
 // CHECK:    call <2 x double> @llvm.x86.sse2.max.pd(<2 x double> %{{.*}}, <2 x double> %{{.*}})
 // CHECK:    shufflevector <2 x double> %{{.*}}, <2 x double>  %{{.*}}, <2 x i32> <i32 1, i32 0>
 // CHECK:    call <2 x double> @llvm.x86.sse2.max.pd(<2 x double> %{{.*}}, <2 x double> %{{.*}})
 // CHECK:    extractelement <2 x double> %{{.*}}, i32 0
   return _mm512_mask_reduce_max_pd(__M, __W); 
 }
 
 long long test_mm512_mask_reduce_min_epi64(__mmask8 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_min_epi64(
 // CHECK:    bitcast i8 %{{.*}} to <8 x i1>
 // CHECK:    select <8 x i1> %{{.*}}, <8 x i64> %{{.*}}, <8 x i64> %{{.*}}
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
-// CHECK:    call <8 x i64> @llvm.smin.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 5>
-// CHECK:    call <8 x i64> @llvm.smin.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
-// CHECK:    call <8 x i64> @llvm.smin.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    extractelement <8 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> %{{.*}})
   return _mm512_mask_reduce_min_epi64(__M, __W); 
 }
 
 unsigned long long test_mm512_mask_reduce_min_epu64(__mmask8 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_min_epu64(
 // CHECK:    bitcast i8 %{{.*}} to <8 x i1>
 // CHECK:    select <8 x i1> %{{.*}}, <8 x i64> %{{.*}}, <8 x i64> %{{.*}}
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
-// CHECK:    call <8 x i64> @llvm.umin.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 5>
-// CHECK:    call <8 x i64> @llvm.umin.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> %{{.*}}, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
-// CHECK:    call <8 x i64> @llvm.umin.v8i64(<8 x i64> %{{.*}}, <8 x i64> %{{.*}})
-// CHECK:    extractelement <8 x i64> %{{.*}}, i32 0
+// CHECK:    call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> %{{.*}})
   return _mm512_mask_reduce_min_epu64(__M, __W); 
 }
 
 double test_mm512_mask_reduce_min_pd(__mmask8 __M, __m512d __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_min_pd(
 // CHECK:    bitcast i8 %{{.*}} to <8 x i1>
 // CHECK:    select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
 // CHECK:    call <4 x double> @llvm.x86.avx.min.pd.256(<4 x double> %{{.*}}, <4 x double> %{{.*}})
 // CHECK:    shufflevector <4 x double> %{{.*}}, <4 x double> undef, <2 x i32> <i32 0, i32 1>
 // CHECK:    shufflevector <4 x double> %{{.*}}, <4 x double> undef, <2 x i32> <i32 2, i32 3>
 // CHECK:    call <2 x double> @llvm.x86.sse2.min.pd(<2 x double> %{{.*}}, <2 x double> %{{.*}})
 // CHECK:    shufflevector <2 x double> %{{.*}}, <2 x double>  %{{.*}}, <2 x i32> <i32 1, i32 0>
 // CHECK:    call <2 x double> @llvm.x86.sse2.min.pd(<2 x double> %{{.*}}, <2 x double> %{{.*}})
 // CHECK:    extractelement <2 x double> %{{.*}}, i32 0
   return _mm512_mask_reduce_min_pd(__M, __W); 
 }
 
 int test_mm512_reduce_max_epi32(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_max_epi32(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    call <8 x i32> @llvm.smax.v8i32(<8 x i32> %{{.*}}, <8 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    call <4 x i32> @llvm.smax.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    call <4 x i32> @llvm.smax.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    call <4 x i32> @llvm.smax.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.smax.v16i32(<16 x i32> %{{.*}})
   return _mm512_reduce_max_epi32(__W);
 }
 
 unsigned int test_mm512_reduce_max_epu32(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_max_epu32(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    call <8 x i32> @llvm.umax.v8i32(<8 x i32> %{{.*}}, <8 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    call <4 x i32> @llvm.umax.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    call <4 x i32> @llvm.umax.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    call <4 x i32> @llvm.umax.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.umax.v16i32(<16 x i32> %{{.*}})
   return _mm512_reduce_max_epu32(__W);
 }
 
 float test_mm512_reduce_max_ps(__m512 __W){
 // CHECK-LABEL: define float @test_mm512_reduce_max_ps(
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
 // CHECK:    call <8 x float> @llvm.x86.avx.max.ps.256(<8 x float> %{{.*}}, <8 x float> %{{.*}})
 // CHECK:    shufflevector <8 x float> %{{.*}}, <8 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 // CHECK:    shufflevector <8 x float> %{{.*}}, <8 x float> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
 // CHECK:    call <4 x float> @llvm.x86.sse.max.ps(<4 x float> %{{.*}}, <4 x float> %{{.*}})
 // CHECK:    shufflevector <4 x float> %{{.*}}, <4 x float> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
 // CHECK:    call <4 x float> @llvm.x86.sse.max.ps(<4 x float> %{{.*}}, <4 x float> %{{.*}})
 // CHECK:    shufflevector <4 x float> %{{.*}}, <4 x float> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
 // CHECK:    call <4 x float> @llvm.x86.sse.max.ps(<4 x float> %{{.*}}, <4 x float> %{{.*}})
 // CHECK:    extractelement <4 x float> %{{.*}}, i32 0
   return _mm512_reduce_max_ps(__W); 
 }
 
 int test_mm512_reduce_min_epi32(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_min_epi32(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    call <8 x i32> @llvm.smin.v8i32(<8 x i32> %{{.*}}, <8 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    call <4 x i32> @llvm.smin.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    call <4 x i32> @llvm.smin.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    call <4 x i32> @llvm.smin.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.smin.v16i32(<16 x i32> %{{.*}})
   return _mm512_reduce_min_epi32(__W);
 }
 
 unsigned int test_mm512_reduce_min_epu32(__m512i __W){
 // CHECK-LABEL: @test_mm512_reduce_min_epu32(
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    call <8 x i32> @llvm.umin.v8i32(<8 x i32> %{{.*}}, <8 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    call <4 x i32> @llvm.umin.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    call <4 x i32> @llvm.umin.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    call <4 x i32> @llvm.umin.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.umin.v16i32(<16 x i32> %{{.*}})
   return _mm512_reduce_min_epu32(__W);
 }
 
 float test_mm512_reduce_min_ps(__m512 __W){
 // CHECK-LABEL: define float @test_mm512_reduce_min_ps(
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
 // CHECK:    call <8 x float> @llvm.x86.avx.min.ps.256(<8 x float> %{{.*}}, <8 x float> %{{.*}})
 // CHECK:    shufflevector <8 x float> %{{.*}}, <8 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 // CHECK:    shufflevector <8 x float> %{{.*}}, <8 x float> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
 // CHECK:    call <4 x float> @llvm.x86.sse.min.ps(<4 x float> %{{.*}}, <4 x float> %{{.*}})
 // CHECK:    shufflevector <4 x float> %{{.*}}, <4 x float> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
 // CHECK:    call <4 x float> @llvm.x86.sse.min.ps(<4 x float> %{{.*}}, <4 x float> %{{.*}})
 // CHECK:    shufflevector <4 x float> %{{.*}}, <4 x float> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
 // CHECK:    call <4 x float> @llvm.x86.sse.min.ps(<4 x float> %{{.*}}, <4 x float> %{{.*}})
 // CHECK:    extractelement <4 x float> %{{.*}}, i32 0
   return _mm512_reduce_min_ps(__W); 
 }
 
 int test_mm512_mask_reduce_max_epi32(__mmask16 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_max_epi32(
 // CHECK:    bitcast i16 %{{.*}} to <16 x i1>
 // CHECK:    select <16 x i1> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> %{{.*}}
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    call <8 x i32> @llvm.smax.v8i32(<8 x i32> %{{.*}}, <8 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    call <4 x i32> @llvm.smax.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    call <4 x i32> @llvm.smax.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    call <4 x i32> @llvm.smax.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.smax.v16i32(<16 x i32> %{{.*}})
   return _mm512_mask_reduce_max_epi32(__M, __W); 
 }
 
 unsigned int test_mm512_mask_reduce_max_epu32(__mmask16 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_max_epu32(
 // CHECK:    bitcast i16 %{{.*}} to <16 x i1>
 // CHECK:    select <16 x i1> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> %{{.*}}
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    call <8 x i32> @llvm.umax.v8i32(<8 x i32> %{{.*}}, <8 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    call <4 x i32> @llvm.umax.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    call <4 x i32> @llvm.umax.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    call <4 x i32> @llvm.umax.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.umax.v16i32(<16 x i32> %{{.*}})
   return _mm512_mask_reduce_max_epu32(__M, __W); 
 }
 
 float test_mm512_mask_reduce_max_ps(__mmask16 __M, __m512 __W){
 // CHECK-LABEL: define float @test_mm512_mask_reduce_max_ps(
 // CHECK:    bitcast i16 %{{.*}} to <16 x i1>
 // CHECK:    select <16 x i1> %{{.*}}, <16 x float> %{{.*}}, <16 x float> %{{.*}}
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>

 // CHECK:    extractelement <4 x float> %{{.*}}, i32 0
   return _mm512_mask_reduce_max_ps(__M, __W); 
 }
 
 int test_mm512_mask_reduce_min_epi32(__mmask16 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_min_epi32(
 // CHECK:    bitcast i16 %{{.*}} to <16 x i1>
 // CHECK:    select <16 x i1> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> %{{.*}}
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    call <8 x i32> @llvm.smin.v8i32(<8 x i32> %{{.*}}, <8 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    call <4 x i32> @llvm.smin.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    call <4 x i32> @llvm.smin.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    call <4 x i32> @llvm.smin.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.smin.v16i32(<16 x i32> %{{.*}})
   return _mm512_mask_reduce_min_epi32(__M, __W); 
 }
 
 unsigned int test_mm512_mask_reduce_min_epu32(__mmask16 __M, __m512i __W){
 // CHECK-LABEL: @test_mm512_mask_reduce_min_epu32(
 // CHECK:    bitcast i16 %{{.*}} to <16 x i1>
 // CHECK:    select <16 x i1> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> %{{.*}}
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-// CHECK:    shufflevector <8 x i64> %{{.*}}, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-// CHECK:    call <8 x i32> @llvm.umin.v8i32(<8 x i32> %{{.*}}, <8 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
-// CHECK:    shufflevector <4 x i64> %{{.*}}, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
-// CHECK:    call <4 x i32> @llvm.umin.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-// CHECK:    call <4 x i32> @llvm.umin.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
-// CHECK:    call <4 x i32> @llvm.umin.v4i32(<4 x i32> %{{.*}}, <4 x i32> %{{.*}})
-// CHECK:    extractelement <4 x i32> %{{.*}}, i32 0
+// CHECK:    call i32 @llvm.vector.reduce.umin.v16i32(<16 x i32> %{{.*}})
   return _mm512_mask_reduce_min_epu32(__M, __W); 
 }
 
 float test_mm512_mask_reduce_min_ps(__mmask16 __M, __m512 __W){
 // CHECK-LABEL: define float @test_mm512_mask_reduce_min_ps(
 // CHECK:    bitcast i16 %{{.*}} to <16 x i1>
 // CHECK:    select <16 x i1> %{{.*}}, <16 x float> %{{.*}}, <16 x float> %{{.*}}
 // CHECK:    shufflevector <8 x double> %{{.*}}, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>