[X86] Add builtins for pshufd, pshuflw, and pshufhw to enable target feature and immediate range checking.

llvm-svn: 334265

13 files changed, 105 insertions, 139 deletions

diff --git a/clang/include/clang/Basic/BuiltinsX86.def b/clang/include/clang/Basic/BuiltinsX86.def
index c63289e25d5..3466e9b0358 100644
--- a/clang/include/clang/Basic/BuiltinsX86.def
+++ b/clang/include/clang/Basic/BuiltinsX86.def
@@ -321,6 +321,9 @@ TARGET_BUILTIN(__builtin_ia32_maskmovdqu, "vV16cV16cc*", "n", "sse2")
 TARGET_BUILTIN(__builtin_ia32_movmskpd, "iV2d", "nc", "sse2")
 TARGET_BUILTIN(__builtin_ia32_pmovmskb128, "iV16c", "nc", "sse2")
 TARGET_BUILTIN(__builtin_ia32_movnti, "vi*i", "n", "sse2")
+TARGET_BUILTIN(__builtin_ia32_pshufd, "V4iV4iIi", "nc", "sse2")
+TARGET_BUILTIN(__builtin_ia32_pshuflw, "V8sV8sIi", "nc", "sse2")
+TARGET_BUILTIN(__builtin_ia32_pshufhw, "V8sV8sIi", "nc", "sse2")
 TARGET_BUILTIN(__builtin_ia32_psadbw128, "V2LLiV16cV16c", "nc", "sse2")
 TARGET_BUILTIN(__builtin_ia32_sqrtpd, "V2dV2d", "nc", "sse2")
 TARGET_BUILTIN(__builtin_ia32_sqrtsd, "V2dV2d", "nc", "sse2")
@@ -598,6 +601,9 @@ TARGET_BUILTIN(__builtin_ia32_pmulhw256, "V16sV16sV16s", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_pmuludq256, "V4LLiV8iV8i", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psadbw256, "V4LLiV32cV32c", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_pshufb256, "V32cV32cV32c", "nc", "avx2")
+TARGET_BUILTIN(__builtin_ia32_pshufd256, "V8iV8iIi", "nc", "avx2")
+TARGET_BUILTIN(__builtin_ia32_pshuflw256, "V16sV16sIi", "nc", "avx2")
+TARGET_BUILTIN(__builtin_ia32_pshufhw256, "V16sV16sIi", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psignb256, "V32cV32cV32c", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psignw256, "V16sV16sV16s", "nc", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psignd256, "V8iV8iV8i", "nc", "avx2")
@@ -1363,6 +1369,8 @@ TARGET_BUILTIN(__builtin_ia32_prorvd128_mask, "V4iV4iV4iV4iUc", "nc", "avx512vl"
 TARGET_BUILTIN(__builtin_ia32_prorvd256_mask, "V8iV8iV8iV8iUc", "nc", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_prorvq128_mask, "V2LLiV2LLiV2LLiV2LLiUc", "nc", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_prorvq256_mask, "V4LLiV4LLiV4LLiV4LLiUc", "nc", "avx512vl")
+TARGET_BUILTIN(__builtin_ia32_pshufhw512, "V32sV32sIi", "nc ", "avx512bw")
+TARGET_BUILTIN(__builtin_ia32_pshuflw512, "V32sV32sIi", "nc ", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_psllv32hi, "V32sV32sV32s", "nc", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_psllw512, "V32sV32sV8s", "nc", "avx512bw")
 TARGET_BUILTIN(__builtin_ia32_psllwi512, "V32sV32si", "nc", "avx512bw")
@@ -1737,6 +1745,7 @@ TARGET_BUILTIN(__builtin_ia32_compresssf512_mask, "V16fV16fV16fUs", "nc", "avx51
 TARGET_BUILTIN(__builtin_ia32_compresssi512_mask, "V16iV16iV16iUs", "nc", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_cmpsd_mask, "UcV2dV2dIiUcIi", "nc", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_cmpss_mask, "UcV4fV4fIiUcIi", "nc", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_pshufd512, "V16iV16iIi", "nc", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_expanddf512_mask, "V8dV8dV8dUc", "nc", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_expanddi512_mask, "V8LLiV8LLiV8LLiUc", "nc", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_expandhi512_mask, "V32sV32sV32sUi", "nc", "avx512vbmi2")
diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 025b34e809c..741f36b095d 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -9325,6 +9325,57 @@ Value *CodeGenFunction::EmitX86BuiltinExpr(unsigned BuiltinID,
                                        makeArrayRef(Indices, NumElts),
                                        "blend");
   }
+  case X86::BI__builtin_ia32_pshuflw:
+  case X86::BI__builtin_ia32_pshuflw256:
+  case X86::BI__builtin_ia32_pshuflw512: {
+    uint32_t Imm = cast<llvm::ConstantInt>(Ops[1])->getZExtValue();
+    llvm::Type *Ty = Ops[0]->getType();
+    unsigned NumElts = Ty->getVectorNumElements();
+
+    // Splat the 8-bits of immediate 4 times to help the loop wrap around.
+    Imm = (Imm & 0xff) * 0x01010101;
+
+    uint32_t Indices[32];
+    for (unsigned l = 0; l != NumElts; l += 8) {
+      for (unsigned i = 0; i != 4; ++i) {
+        Indices[l + i] = l + (Imm & 3);
+        Imm >>= 2;
+      }
+      for (unsigned i = 4; i != 8; ++i)
+        Indices[l + i] = l + i;
+    }
+
+    return Builder.CreateShuffleVector(Ops[0], UndefValue::get(Ty),
+                                       makeArrayRef(Indices, NumElts),
+                                       "pshuflw");
+  }
+  case X86::BI__builtin_ia32_pshufhw:
+  case X86::BI__builtin_ia32_pshufhw256:
+  case X86::BI__builtin_ia32_pshufhw512: {
+    uint32_t Imm = cast<llvm::ConstantInt>(Ops[1])->getZExtValue();
+    llvm::Type *Ty = Ops[0]->getType();
+    unsigned NumElts = Ty->getVectorNumElements();
+
+    // Splat the 8-bits of immediate 4 times to help the loop wrap around.
+    Imm = (Imm & 0xff) * 0x01010101;
+
+    uint32_t Indices[32];
+    for (unsigned l = 0; l != NumElts; l += 8) {
+      for (unsigned i = 0; i != 4; ++i)
+        Indices[l + i] = l + i;
+      for (unsigned i = 4; i != 8; ++i) {
+        Indices[l + i] = l + 4 + (Imm & 3);
+        Imm >>= 2;
+      }
+    }
+
+    return Builder.CreateShuffleVector(Ops[0], UndefValue::get(Ty),
+                                       makeArrayRef(Indices, NumElts),
+                                       "pshufhw");
+  }
+  case X86::BI__builtin_ia32_pshufd:
+  case X86::BI__builtin_ia32_pshufd256:
+  case X86::BI__builtin_ia32_pshufd512:
   case X86::BI__builtin_ia32_vpermilpd:
   case X86::BI__builtin_ia32_vpermilps:
   case X86::BI__builtin_ia32_vpermilpd256:
diff --git a/clang/lib/Headers/avx2intrin.h b/clang/lib/Headers/avx2intrin.h
index 3867af08ccc..e7fdd2b1e59 100644
--- a/clang/lib/Headers/avx2intrin.h
+++ b/clang/lib/Headers/avx2intrin.h
@@ -488,44 +488,13 @@ _mm256_shuffle_epi8(__m256i __a, __m256i __b)
 }
 
 #define _mm256_shuffle_epi32(a, imm) \
-  (__m256i)__builtin_shufflevector((__v8si)(__m256i)(a), \
-                                   (__v8si)_mm256_undefined_si256(), \
-                                   0 + (((imm) >> 0) & 0x3), \
-                                   0 + (((imm) >> 2) & 0x3), \
-                                   0 + (((imm) >> 4) & 0x3), \
-                                   0 + (((imm) >> 6) & 0x3), \
-                                   4 + (((imm) >> 0) & 0x3), \
-                                   4 + (((imm) >> 2) & 0x3), \
-                                   4 + (((imm) >> 4) & 0x3), \
-                                   4 + (((imm) >> 6) & 0x3))
+  (__m256i)__builtin_ia32_pshufd256((__v8si)(__m256i)(a), (int)(imm))
 
 #define _mm256_shufflehi_epi16(a, imm) \
-  (__m256i)__builtin_shufflevector((__v16hi)(__m256i)(a), \
-                                   (__v16hi)_mm256_undefined_si256(), \
-                                   0, 1, 2, 3, \
-                                   4  + (((imm) >> 0) & 0x3), \
-                                   4  + (((imm) >> 2) & 0x3), \
-                                   4  + (((imm) >> 4) & 0x3), \
-                                   4  + (((imm) >> 6) & 0x3), \
-                                   8, 9, 10, 11, \
-                                   12 + (((imm) >> 0) & 0x3), \
-                                   12 + (((imm) >> 2) & 0x3), \
-                                   12 + (((imm) >> 4) & 0x3), \
-                                   12 + (((imm) >> 6) & 0x3))
+  (__m256i)__builtin_ia32_pshufhw256((__v16hi)(__m256i)(a), (int)(imm))
 
 #define _mm256_shufflelo_epi16(a, imm) \
-  (__m256i)__builtin_shufflevector((__v16hi)(__m256i)(a), \
-                                   (__v16hi)_mm256_undefined_si256(), \
-                                   0 + (((imm) >> 0) & 0x3), \
-                                   0 + (((imm) >> 2) & 0x3), \
-                                   0 + (((imm) >> 4) & 0x3), \
-                                   0 + (((imm) >> 6) & 0x3), \
-                                   4, 5, 6, 7, \
-                                   8 + (((imm) >> 0) & 0x3), \
-                                   8 + (((imm) >> 2) & 0x3), \
-                                   8 + (((imm) >> 4) & 0x3), \
-                                   8 + (((imm) >> 6) & 0x3), \
-                                   12, 13, 14, 15)
+  (__m256i)__builtin_ia32_pshuflw256((__v16hi)(__m256i)(a), (int)(imm))
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_sign_epi8(__m256i __a, __m256i __b)
diff --git a/clang/lib/Headers/avx512bwintrin.h b/clang/lib/Headers/avx512bwintrin.h
index aabaabd648d..0e21a57f311 100644
--- a/clang/lib/Headers/avx512bwintrin.h
+++ b/clang/lib/Headers/avx512bwintrin.h
@@ -1287,28 +1287,7 @@ _mm512_maskz_cvtepu8_epi16(__mmask32 __U, __m256i __A)
 
 
 #define _mm512_shufflehi_epi16(A, imm) \
-  (__m512i)__builtin_shufflevector((__v32hi)(__m512i)(A), \
-                                   (__v32hi)_mm512_undefined_epi32(), \
-                                   0, 1, 2, 3, \
-                                   4  + (((imm) >> 0) & 0x3), \
-                                   4  + (((imm) >> 2) & 0x3), \
-                                   4  + (((imm) >> 4) & 0x3), \
-                                   4  + (((imm) >> 6) & 0x3), \
-                                   8, 9, 10, 11, \
-                                   12 + (((imm) >> 0) & 0x3), \
-                                   12 + (((imm) >> 2) & 0x3), \
-                                   12 + (((imm) >> 4) & 0x3), \
-                                   12 + (((imm) >> 6) & 0x3), \
-                                   16, 17, 18, 19, \
-                                   20 + (((imm) >> 0) & 0x3), \
-                                   20 + (((imm) >> 2) & 0x3), \
-                                   20 + (((imm) >> 4) & 0x3), \
-                                   20 + (((imm) >> 6) & 0x3), \
-                                   24, 25, 26, 27, \
-                                   28 + (((imm) >> 0) & 0x3), \
-                                   28 + (((imm) >> 2) & 0x3), \
-                                   28 + (((imm) >> 4) & 0x3), \
-                                   28 + (((imm) >> 6) & 0x3))
+  (__m512i)__builtin_ia32_pshufhw512((__v32hi)(__m512i)(A), (int)(imm))
 
 #define _mm512_mask_shufflehi_epi16(W, U, A, imm) \
   (__m512i)__builtin_ia32_selectw_512((__mmask32)(U), \
@@ -1323,28 +1302,7 @@ _mm512_maskz_cvtepu8_epi16(__mmask32 __U, __m256i __A)
                                       (__v32hi)_mm512_setzero_si512())
 
 #define _mm512_shufflelo_epi16(A, imm) \
-  (__m512i)__builtin_shufflevector((__v32hi)(__m512i)(A), \
-                                   (__v32hi)_mm512_undefined_epi32(), \
-                                   0 + (((imm) >> 0) & 0x3), \
-                                   0 + (((imm) >> 2) & 0x3), \
-                                   0 + (((imm) >> 4) & 0x3), \
-                                   0 + (((imm) >> 6) & 0x3), \
-                                   4, 5, 6, 7, \
-                                   8 + (((imm) >> 0) & 0x3), \
-                                   8 + (((imm) >> 2) & 0x3), \
-                                   8 + (((imm) >> 4) & 0x3), \
-                                   8 + (((imm) >> 6) & 0x3), \
-                                   12, 13, 14, 15, \
-                                   16 + (((imm) >> 0) & 0x3), \
-                                   16 + (((imm) >> 2) & 0x3), \
-                                   16 + (((imm) >> 4) & 0x3), \
-                                   16 + (((imm) >> 6) & 0x3), \
-                                   20, 21, 22, 23, \
-                                   24 + (((imm) >> 0) & 0x3), \
-                                   24 + (((imm) >> 2) & 0x3), \
-                                   24 + (((imm) >> 4) & 0x3), \
-                                   24 + (((imm) >> 6) & 0x3), \
-                                   28, 29, 30, 31)
+  (__m512i)__builtin_ia32_pshuflw512((__v32hi)(__m512i)(A), (int)(imm))
 
 
 #define _mm512_mask_shufflelo_epi16(W, U, A, imm) \
diff --git a/clang/lib/Headers/avx512fintrin.h b/clang/lib/Headers/avx512fintrin.h
index 4ae235e6330..fe730ac27f2 100644
--- a/clang/lib/Headers/avx512fintrin.h
+++ b/clang/lib/Headers/avx512fintrin.h
@@ -8849,24 +8849,7 @@ _mm_maskz_load_sd (__mmask8 __U, const double* __A)
 }
 
 #define _mm512_shuffle_epi32(A, I) \
-  (__m512i)__builtin_shufflevector((__v16si)(__m512i)(A), \
-                                   (__v16si)_mm512_undefined_epi32(), \
-                                   0  + (((I) >> 0) & 0x3), \
-                                   0  + (((I) >> 2) & 0x3), \
-                                   0  + (((I) >> 4) & 0x3), \
-                                   0  + (((I) >> 6) & 0x3), \
-                                   4  + (((I) >> 0) & 0x3), \
-                                   4  + (((I) >> 2) & 0x3), \
-                                   4  + (((I) >> 4) & 0x3), \
-                                   4  + (((I) >> 6) & 0x3), \
-                                   8  + (((I) >> 0) & 0x3), \
-                                   8  + (((I) >> 2) & 0x3), \
-                                   8  + (((I) >> 4) & 0x3), \
-                                   8  + (((I) >> 6) & 0x3), \
-                                   12 + (((I) >> 0) & 0x3), \
-                                   12 + (((I) >> 2) & 0x3), \
-                                   12 + (((I) >> 4) & 0x3), \
-                                   12 + (((I) >> 6) & 0x3))
+  (__m512i)__builtin_ia32_pshufd512((__v16si)(__m512i)(A), (int)(I))
 
 #define _mm512_mask_shuffle_epi32(W, U, A, I) \
   (__m512i)__builtin_ia32_selectd_512((__mmask16)(U), \
diff --git a/clang/lib/Headers/emmintrin.h b/clang/lib/Headers/emmintrin.h
index ad836b1e637..8942790d5c5 100644
--- a/clang/lib/Headers/emmintrin.h
+++ b/clang/lib/Headers/emmintrin.h
@@ -4343,10 +4343,7 @@ _mm_movemask_epi8(__m128i __a)
 ///    11: assign values from bits [127:96] of \a a.
 /// \returns A 128-bit integer vector containing the shuffled values.
 #define _mm_shuffle_epi32(a, imm) \
-  (__m128i)__builtin_shufflevector((__v4si)(__m128i)(a), \
-                                   (__v4si)_mm_undefined_si128(), \
-                                   ((imm) >> 0) & 0x3, ((imm) >> 2) & 0x3, \
-                                   ((imm) >> 4) & 0x3, ((imm) >> 6) & 0x3)
+  (__m128i)__builtin_ia32_pshufd((__v4si)(__m128i)(a), (int)(imm))
 
 /// Constructs a 128-bit integer vector by shuffling four lower 16-bit
 ///    elements of a 128-bit integer vector of [8 x i16], using the immediate
@@ -4376,11 +4373,7 @@ _mm_movemask_epi8(__m128i __a)
 ///    11: assign values from bits [63:48] of \a a. \n
 /// \returns A 128-bit integer vector containing the shuffled values.
 #define _mm_shufflelo_epi16(a, imm) \
-  (__m128i)__builtin_shufflevector((__v8hi)(__m128i)(a), \
-                                   (__v8hi)_mm_undefined_si128(), \
-                                   ((imm) >> 0) & 0x3, ((imm) >> 2) & 0x3, \
-                                   ((imm) >> 4) & 0x3, ((imm) >> 6) & 0x3, \
-                                   4, 5, 6, 7)
+  (__m128i)__builtin_ia32_pshuflw((__v8hi)(__m128i)(a), (int)(imm))
 
 /// Constructs a 128-bit integer vector by shuffling four upper 16-bit
 ///    elements of a 128-bit integer vector of [8 x i16], using the immediate
@@ -4410,13 +4403,7 @@ _mm_movemask_epi8(__m128i __a)
 ///    11: assign values from bits [127:112] of \a a. \n
 /// \returns A 128-bit integer vector containing the shuffled values.
 #define _mm_shufflehi_epi16(a, imm) \
-  (__m128i)__builtin_shufflevector((__v8hi)(__m128i)(a), \
-                                   (__v8hi)_mm_undefined_si128(), \
-                                   0, 1, 2, 3, \
-                                   4 + (((imm) >> 0) & 0x3), \
-                                   4 + (((imm) >> 2) & 0x3), \
-                                   4 + (((imm) >> 4) & 0x3), \
-                                   4 + (((imm) >> 6) & 0x3))
+  (__m128i)__builtin_ia32_pshufhw((__v8hi)(__m128i)(a), (int)(imm))
 
 /// Unpacks the high-order (index 8-15) values from two 128-bit vectors
 ///    of [16 x i8] and interleaves them into a 128-bit vector of [16 x i8].
diff --git a/clang/lib/Sema/SemaChecking.cpp b/clang/lib/Sema/SemaChecking.cpp
index 4f5405935e2..020d82edf72 100644
--- a/clang/lib/Sema/SemaChecking.cpp
+++ b/clang/lib/Sema/SemaChecking.cpp
@@ -2759,6 +2759,15 @@ bool Sema::CheckX86BuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
   case X86::BI__builtin_ia32_vpermilps256:
   case X86::BI__builtin_ia32_vpermilpd512:
   case X86::BI__builtin_ia32_vpermilps512:
+  case X86::BI__builtin_ia32_pshufd:
+  case X86::BI__builtin_ia32_pshufd256:
+  case X86::BI__builtin_ia32_pshufd512:
+  case X86::BI__builtin_ia32_pshufhw:
+  case X86::BI__builtin_ia32_pshufhw256:
+  case X86::BI__builtin_ia32_pshufhw512:
+  case X86::BI__builtin_ia32_pshuflw:
+  case X86::BI__builtin_ia32_pshuflw256:
+  case X86::BI__builtin_ia32_pshuflw512:
   case X86::BI__builtin_ia32_vcvtps2ph:
   case X86::BI__builtin_ia32_vcvtps2ph_mask:
   case X86::BI__builtin_ia32_vcvtps2ph256:
diff --git a/clang/test/CodeGen/avx2-builtins.c b/clang/test/CodeGen/avx2-builtins.c
index 72b4349f6e8..99c304109c6 100644
--- a/clang/test/CodeGen/avx2-builtins.c
+++ b/clang/test/CodeGen/avx2-builtins.c
@@ -955,19 +955,19 @@ __m256i test_mm256_shuffle_epi8(__m256i a, __m256i b) {
 
 __m256i test_mm256_shuffle_epi32(__m256i a) {
   // CHECK-LABEL: test_mm256_shuffle_epi32
-  // CHECK: shufflevector <8 x i32> %{{.*}}, <8 x i32> %{{.*}}, <8 x i32> <i32 3, i32 3, i32 0, i32 0, i32 7, i32 7, i32 4, i32 4>
+  // CHECK: shufflevector <8 x i32> %{{.*}}, <8 x i32> undef, <8 x i32> <i32 3, i32 3, i32 0, i32 0, i32 7, i32 7, i32 4, i32 4>
   return _mm256_shuffle_epi32(a, 15);
 }
 
 __m256i test_mm256_shufflehi_epi16(__m256i a) {
   // CHECK-LABEL: test_mm256_shufflehi_epi16
-  // CHECK: shufflevector <16 x i16> %{{.*}}, <16 x i16> %{{.*}}, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 7, i32 6, i32 6, i32 5, i32 8, i32 9, i32 10, i32 11, i32 15, i32 14, i32 14, i32 13>
+  // CHECK: shufflevector <16 x i16> %{{.*}}, <16 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 7, i32 6, i32 6, i32 5, i32 8, i32 9, i32 10, i32 11, i32 15, i32 14, i32 14, i32 13>
   return _mm256_shufflehi_epi16(a, 107);
 }
 
 __m256i test_mm256_shufflelo_epi16(__m256i a) {
   // CHECK-LABEL: test_mm256_shufflelo_epi16
-  // CHECK: shufflevector <16 x i16> %{{.*}}, <16 x i16> %{{.*}}, <16 x i32> <i32 3, i32 0, i32 1, i32 1, i32 4, i32 5, i32 6, i32 7, i32 11, i32 8, i32 9, i32 9, i32 12, i32 13, i32 14, i32 15>
+  // CHECK: shufflevector <16 x i16> %{{.*}}, <16 x i16> undef, <16 x i32> <i32 3, i32 0, i32 1, i32 1, i32 4, i32 5, i32 6, i32 7, i32 11, i32 8, i32 9, i32 9, i32 12, i32 13, i32 14, i32 15>
   return _mm256_shufflelo_epi16(a, 83);
 }
 
diff --git a/clang/test/CodeGen/avx512bw-builtins.c b/clang/test/CodeGen/avx512bw-builtins.c
index 07569f53c7e..6803652a794 100644
--- a/clang/test/CodeGen/avx512bw-builtins.c
+++ b/clang/test/CodeGen/avx512bw-builtins.c
@@ -1250,40 +1250,40 @@ __m512i test_mm512_maskz_cvtepu8_epi16(__mmask32 __U, __m256i __A) {
 
 __m512i test_mm512_shufflehi_epi16(__m512i __A) {
   // CHECK-LABEL: @test_mm512_shufflehi_epi16
-  // CHECK: shufflevector <32 x i16> %{{.*}}, <32 x i16> %{{.*}}, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 5, i32 4, i32 4, i32 8, i32 9, i32 10, i32 11, i32 13, i32 13, i32 12, i32 12, i32 16, i32 17, i32 18, i32 19, i32 21, i32 21, i32 20, i32 20, i32 24, i32 25, i32 26, i32 27, i32 29, i32 29, i32 28, i32 28>
+  // CHECK: shufflevector <32 x i16> %{{.*}}, <32 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 5, i32 4, i32 4, i32 8, i32 9, i32 10, i32 11, i32 13, i32 13, i32 12, i32 12, i32 16, i32 17, i32 18, i32 19, i32 21, i32 21, i32 20, i32 20, i32 24, i32 25, i32 26, i32 27, i32 29, i32 29, i32 28, i32 28>
   return _mm512_shufflehi_epi16(__A, 5); 
 }
 
 __m512i test_mm512_mask_shufflehi_epi16(__m512i __W, __mmask32 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_mask_shufflehi_epi16
-  // CHECK: shufflevector <32 x i16> %{{.*}}, <32 x i16> %{{.*}}, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 5, i32 4, i32 4, i32 8, i32 9, i32 10, i32 11, i32 13, i32 13, i32 12, i32 12, i32 16, i32 17, i32 18, i32 19, i32 21, i32 21, i32 20, i32 20, i32 24, i32 25, i32 26, i32 27, i32 29, i32 29, i32 28, i32 28>
+  // CHECK: shufflevector <32 x i16> %{{.*}}, <32 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 5, i32 4, i32 4, i32 8, i32 9, i32 10, i32 11, i32 13, i32 13, i32 12, i32 12, i32 16, i32 17, i32 18, i32 19, i32 21, i32 21, i32 20, i32 20, i32 24, i32 25, i32 26, i32 27, i32 29, i32 29, i32 28, i32 28>
   // CHECK: select <32 x i1> %{{.*}}, <32 x i16> %{{.*}}, <32 x i16> %{{.*}}
   return _mm512_mask_shufflehi_epi16(__W, __U, __A, 5); 
 }
 
 __m512i test_mm512_maskz_shufflehi_epi16(__mmask32 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_maskz_shufflehi_epi16
-  // CHECK: shufflevector <32 x i16> %{{.*}}, <32 x i16> %{{.*}}, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 5, i32 4, i32 4, i32 8, i32 9, i32 10, i32 11, i32 13, i32 13, i32 12, i32 12, i32 16, i32 17, i32 18, i32 19, i32 21, i32 21, i32 20, i32 20, i32 24, i32 25, i32 26, i32 27, i32 29, i32 29, i32 28, i32 28>
+  // CHECK: shufflevector <32 x i16> %{{.*}}, <32 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 5, i32 4, i32 4, i32 8, i32 9, i32 10, i32 11, i32 13, i32 13, i32 12, i32 12, i32 16, i32 17, i32 18, i32 19, i32 21, i32 21, i32 20, i32 20, i32 24, i32 25, i32 26, i32 27, i32 29, i32 29, i32 28, i32 28>
   // CHECK: select <32 x i1> %{{.*}}, <32 x i16> %{{.*}}, <32 x i16> %{{.*}}
   return _mm512_maskz_shufflehi_epi16(__U, __A, 5); 
 }
 
 __m512i test_mm512_shufflelo_epi16(__m512i __A) {
   // CHECK-LABEL: @test_mm512_shufflelo_epi16
-  // CHECK: shufflevector <32 x i16> %{{.*}}, <32 x i16> %{{.*}}, <32 x i32> <i32 1, i32 1, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7, i32 9, i32 9, i32 8, i32 8, i32 12, i32 13, i32 14, i32 15, i32 17, i32 17, i32 16, i32 16, i32 20, i32 21, i32 22, i32 23, i32 25, i32 25, i32 24, i32 24, i32 28, i32 29, i32 30, i32 31>
+  // CHECK: shufflevector <32 x i16> %{{.*}}, <32 x i16> undef, <32 x i32> <i32 1, i32 1, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7, i32 9, i32 9, i32 8, i32 8, i32 12, i32 13, i32 14, i32 15, i32 17, i32 17, i32 16, i32 16, i32 20, i32 21, i32 22, i32 23, i32 25, i32 25, i32 24, i32 24, i32 28, i32 29, i32 30, i32 31>
   return _mm512_shufflelo_epi16(__A, 5); 
 }
 
 __m512i test_mm512_mask_shufflelo_epi16(__m512i __W, __mmask32 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_mask_shufflelo_epi16
-  // CHECK: shufflevector <32 x i16> %{{.*}}, <32 x i16> %{{.*}}, <32 x i32> <i32 1, i32 1, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7, i32 9, i32 9, i32 8, i32 8, i32 12, i32 13, i32 14, i32 15, i32 17, i32 17, i32 16, i32 16, i32 20, i32 21, i32 22, i32 23, i32 25, i32 25, i32 24, i32 24, i32 28, i32 29, i32 30, i32 31>
+  // CHECK: shufflevector <32 x i16> %{{.*}}, <32 x i16> undef, <32 x i32> <i32 1, i32 1, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7, i32 9, i32 9, i32 8, i32 8, i32 12, i32 13, i32 14, i32 15, i32 17, i32 17, i32 16, i32 16, i32 20, i32 21, i32 22, i32 23, i32 25, i32 25, i32 24, i32 24, i32 28, i32 29, i32 30, i32 31>
   // CHECK: select <32 x i1> %{{.*}}, <32 x i16> %{{.*}}, <32 x i16> %{{.*}}
   return _mm512_mask_shufflelo_epi16(__W, __U, __A, 5); 
 }
 
 __m512i test_mm512_maskz_shufflelo_epi16(__mmask32 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_maskz_shufflelo_epi16
-  // CHECK: shufflevector <32 x i16> %{{.*}}, <32 x i16> %{{.*}}, <32 x i32> <i32 1, i32 1, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7, i32 9, i32 9, i32 8, i32 8, i32 12, i32 13, i32 14, i32 15, i32 17, i32 17, i32 16, i32 16, i32 20, i32 21, i32 22, i32 23, i32 25, i32 25, i32 24, i32 24, i32 28, i32 29, i32 30, i32 31>
+  // CHECK: shufflevector <32 x i16> %{{.*}}, <32 x i16> undef, <32 x i32> <i32 1, i32 1, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7, i32 9, i32 9, i32 8, i32 8, i32 12, i32 13, i32 14, i32 15, i32 17, i32 17, i32 16, i32 16, i32 20, i32 21, i32 22, i32 23, i32 25, i32 25, i32 24, i32 24, i32 28, i32 29, i32 30, i32 31>
   // CHECK: select <32 x i1> %{{.*}}, <32 x i16> %{{.*}}, <32 x i16> %{{.*}}
   return _mm512_maskz_shufflelo_epi16(__U, __A, 5); 
 }
diff --git a/clang/test/CodeGen/avx512f-builtins.c b/clang/test/CodeGen/avx512f-builtins.c
index 52554ba183e..dd55c43940f 100644
--- a/clang/test/CodeGen/avx512f-builtins.c
+++ b/clang/test/CodeGen/avx512f-builtins.c
@@ -7234,20 +7234,20 @@ __m512 test_mm512_maskz_moveldup_ps(__mmask16 __U, __m512 __A) {
 
 __m512i test_mm512_shuffle_epi32(__m512i __A) {
   // CHECK-LABEL: @test_mm512_shuffle_epi32
-  // CHECK: shufflevector <16 x i32> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> <i32 1, i32 0, i32 0, i32 0, i32 5, i32 4, i32 4, i32 4, i32 9, i32 8, i32 8, i32 8, i32 13, i32 12, i32 12, i32 12>
+  // CHECK: shufflevector <16 x i32> %{{.*}}, <16 x i32> undef, <16 x i32> <i32 1, i32 0, i32 0, i32 0, i32 5, i32 4, i32 4, i32 4, i32 9, i32 8, i32 8, i32 8, i32 13, i32 12, i32 12, i32 12>
   return _mm512_shuffle_epi32(__A, 1); 
 }
 
 __m512i test_mm512_mask_shuffle_epi32(__m512i __W, __mmask16 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_mask_shuffle_epi32
-  // CHECK: shufflevector <16 x i32> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> <i32 1, i32 0, i32 0, i32 0, i32 5, i32 4, i32 4, i32 4, i32 9, i32 8, i32 8, i32 8, i32 13, i32 12, i32 12, i32 12>
+  // CHECK: shufflevector <16 x i32> %{{.*}}, <16 x i32> undef, <16 x i32> <i32 1, i32 0, i32 0, i32 0, i32 5, i32 4, i32 4, i32 4, i32 9, i32 8, i32 8, i32 8, i32 13, i32 12, i32 12, i32 12>
   // CHECK: select <16 x i1> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> %{{.*}}
   return _mm512_mask_shuffle_epi32(__W, __U, __A, 1); 
 }
 
 __m512i test_mm512_maskz_shuffle_epi32(__mmask16 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_maskz_shuffle_epi32
-  // CHECK: shufflevector <16 x i32> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> <i32 1, i32 0, i32 0, i32 0, i32 5, i32 4, i32 4, i32 4, i32 9, i32 8, i32 8, i32 8, i32 13, i32 12, i32 12, i32 12>
+  // CHECK: shufflevector <16 x i32> %{{.*}}, <16 x i32> undef, <16 x i32> <i32 1, i32 0, i32 0, i32 0, i32 5, i32 4, i32 4, i32 4, i32 9, i32 8, i32 8, i32 8, i32 13, i32 12, i32 12, i32 12>
   // CHECK: select <16 x i1> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> %{{.*}}
   return _mm512_maskz_shuffle_epi32(__U, __A, 1); 
 }
diff --git a/clang/test/CodeGen/avx512vl-builtins.c b/clang/test/CodeGen/avx512vl-builtins.c
index 1edc25d8494..c234e3b94d8 100644
--- a/clang/test/CodeGen/avx512vl-builtins.c
+++ b/clang/test/CodeGen/avx512vl-builtins.c
@@ -7598,28 +7598,28 @@ __m256 test_mm256_maskz_moveldup_ps(__mmask8 __U, __m256 __A) {
 
 __m128i test_mm_mask_shuffle_epi32(__m128i __W, __mmask8 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_mask_shuffle_epi32
-  // CHECK: shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
+  // CHECK: shufflevector <4 x i32> %{{.*}}, <4 x i32> undef, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
   // CHECK: select <4 x i1> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> %{{.*}}
   return _mm_mask_shuffle_epi32(__W, __U, __A, 1);
 }
 
 __m128i test_mm_maskz_shuffle_epi32(__mmask8 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_maskz_shuffle_epi32
-  // CHECK: shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> <i32 2, i32 0, i32 0, i32 0>
+  // CHECK: shufflevector <4 x i32> %{{.*}}, <4 x i32> undef, <4 x i32> <i32 2, i32 0, i32 0, i32 0>
   // CHECK: select <4 x i1> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> %{{.*}}
   return _mm_maskz_shuffle_epi32(__U, __A, 2);
 }
 
 __m256i test_mm256_mask_shuffle_epi32(__m256i __W, __mmask8 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_mask_shuffle_epi32
-  // CHECK: shufflevector <8 x i32> %{{.*}}, <8 x i32> %{{.*}}, <8 x i32> <i32 2, i32 0, i32 0, i32 0, i32 6, i32 4, i32 4, i32 4>
+  // CHECK: shufflevector <8 x i32> %{{.*}}, <8 x i32> undef, <8 x i32> <i32 2, i32 0, i32 0, i32 0, i32 6, i32 4, i32 4, i32 4>
   // CHECK: select <8 x i1> %{{.*}}, <8 x i32> %{{.*}}, <8 x i32> %{{.*}}
   return _mm256_mask_shuffle_epi32(__W, __U, __A, 2);
 }
 
 __m256i test_mm256_maskz_shuffle_epi32(__mmask8 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_maskz_shuffle_epi32
-  // CHECK: shufflevector <8 x i32> %{{.*}}, <8 x i32> %{{.*}}, <8 x i32> <i32 2, i32 0, i32 0, i32 0, i32 6, i32 4, i32 4, i32 4>
+  // CHECK: shufflevector <8 x i32> %{{.*}}, <8 x i32> undef, <8 x i32> <i32 2, i32 0, i32 0, i32 0, i32 6, i32 4, i32 4, i32 4>
   // CHECK: select <8 x i1> %{{.*}}, <8 x i32> %{{.*}}, <8 x i32> %{{.*}}
   return _mm256_maskz_shuffle_epi32(__U, __A, 2);
 }
diff --git a/clang/test/CodeGen/avx512vlbw-builtins.c b/clang/test/CodeGen/avx512vlbw-builtins.c
index da16abd2a81..9cc58b65057 100644
--- a/clang/test/CodeGen/avx512vlbw-builtins.c
+++ b/clang/test/CodeGen/avx512vlbw-builtins.c
@@ -3013,56 +3013,56 @@ __mmask16 test_mm256_movepi16_mask(__m256i __A) {
 
 __m128i test_mm_mask_shufflehi_epi16(__m128i __W, __mmask32 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_mask_shufflehi_epi16
-  // CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> %{{.*}}, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 5, i32 4, i32 4>
+  // CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 5, i32 4, i32 4>
   // CHECK: select <8 x i1> %{{.*}}, <8 x i16> %{{.*}}, <8 x i16> %{{.*}}
   return _mm_mask_shufflehi_epi16(__W, __U, __A, 5); 
 }
 
 __m128i test_mm_maskz_shufflehi_epi16(__mmask32 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_maskz_shufflehi_epi16
-  // CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> %{{.*}}, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 5, i32 4, i32 4>
+  // CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 5, i32 4, i32 4>
   // CHECK: select <8 x i1> %{{.*}}, <8 x i16> %{{.*}}, <8 x i16> %{{.*}}
   return _mm_maskz_shufflehi_epi16(__U, __A, 5); 
 }
 
 __m128i test_mm_mask_shufflelo_epi16(__m128i __W, __mmask32 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_mask_shufflelo_epi16
-  // CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> %{{.*}}, <8 x i32> <i32 1, i32 1, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7>
+  // CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> undef, <8 x i32> <i32 1, i32 1, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7>
   // CHECK: select <8 x i1> %{{.*}}, <8 x i16> %{{.*}}, <8 x i16> %{{.*}}
   return _mm_mask_shufflelo_epi16(__W, __U, __A, 5); 
 }
 
 __m128i test_mm_maskz_shufflelo_epi16(__mmask32 __U, __m128i __A) {
   // CHECK-LABEL: @test_mm_maskz_shufflelo_epi16
-  // CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> %{{.*}}, <8 x i32> <i32 1, i32 1, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7>
+  // CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> undef, <8 x i32> <i32 1, i32 1, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7>
   // CHECK: select <8 x i1> %{{.*}}, <8 x i16> %{{.*}}, <8 x i16> %{{.*}}
   return _mm_maskz_shufflelo_epi16(__U, __A, 5); 
 }
 
 __m256i test_mm256_mask_shufflehi_epi16(__m256i __W, __mmask32 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_mask_shufflehi_epi16
-  // CHECK: shufflevector <16 x i16> %{{.*}}, <16 x i16> %{{.*}}, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 5, i32 4, i32 4, i32 8, i32 9, i32 10, i32 11, i32 13, i32 13, i32 12, i32 12>
+  // CHECK: shufflevector <16 x i16> %{{.*}}, <16 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 5, i32 4, i32 4, i32 8, i32 9, i32 10, i32 11, i32 13, i32 13, i32 12, i32 12>
   // CHECK: select <16 x i1> %{{.*}}, <16 x i16> %{{.*}}, <16 x i16> %{{.*}}
   return _mm256_mask_shufflehi_epi16(__W, __U, __A, 5); 
 }
 
 __m256i test_mm256_maskz_shufflehi_epi16(__mmask32 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_maskz_shufflehi_epi16
-  // CHECK: shufflevector <16 x i16> %{{.*}}, <16 x i16> %{{.*}}, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 5, i32 4, i32 4, i32 8, i32 9, i32 10, i32 11, i32 13, i32 13, i32 12, i32 12>
+  // CHECK: shufflevector <16 x i16> %{{.*}}, <16 x i16> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 5, i32 4, i32 4, i32 8, i32 9, i32 10, i32 11, i32 13, i32 13, i32 12, i32 12>
   // CHECK: select <16 x i1> %{{.*}}, <16 x i16> %{{.*}}, <16 x i16> %{{.*}}
   return _mm256_maskz_shufflehi_epi16(__U, __A, 5); 
 }
 
 __m256i test_mm256_mask_shufflelo_epi16(__m256i __W, __mmask32 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_mask_shufflelo_epi16
-  // CHECK: shufflevector <16 x i16> %{{.*}}, <16 x i16> %{{.*}}, <16 x i32> <i32 1, i32 1, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7, i32 9, i32 9, i32 8, i32 8, i32 12, i32 13, i32 14, i32 15>
+  // CHECK: shufflevector <16 x i16> %{{.*}}, <16 x i16> undef, <16 x i32> <i32 1, i32 1, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7, i32 9, i32 9, i32 8, i32 8, i32 12, i32 13, i32 14, i32 15>
   // CHECK: select <16 x i1> %{{.*}}, <16 x i16> %{{.*}}, <16 x i16> %{{.*}}
   return _mm256_mask_shufflelo_epi16(__W, __U, __A, 5); 
 }
 
 __m256i test_mm256_maskz_shufflelo_epi16(__mmask32 __U, __m256i __A) {
   // CHECK-LABEL: @test_mm256_maskz_shufflelo_epi16
-  // CHECK: shufflevector <16 x i16> %{{.*}}, <16 x i16> %{{.*}}, <16 x i32> <i32 1, i32 1, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7, i32 9, i32 9, i32 8, i32 8, i32 12, i32 13, i32 14, i32 15>
+  // CHECK: shufflevector <16 x i16> %{{.*}}, <16 x i16> undef, <16 x i32> <i32 1, i32 1, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7, i32 9, i32 9, i32 8, i32 8, i32 12, i32 13, i32 14, i32 15>
   // CHECK: select <16 x i1> %{{.*}}, <16 x i16> %{{.*}}, <16 x i16> %{{.*}}
   return _mm256_maskz_shufflelo_epi16(__U, __A, 5); 
 }
diff --git a/clang/test/CodeGen/sse2-builtins.c b/clang/test/CodeGen/sse2-builtins.c
index 0031215f7c5..0d79aabbafb 100644
--- a/clang/test/CodeGen/sse2-builtins.c
+++ b/clang/test/CodeGen/sse2-builtins.c
@@ -1116,7 +1116,7 @@ __m128i test_mm_setzero_si128() {
 
 __m128i test_mm_shuffle_epi32(__m128i A) {
   // CHECK-LABEL: test_mm_shuffle_epi32
-  // CHECK: shufflevector <4 x i32> %{{.*}}, <4 x i32> %{{.*}}, <4 x i32> zeroinitializer
+  // CHECK: shufflevector <4 x i32> %{{.*}}, <4 x i32> undef, <4 x i32> zeroinitializer
   return _mm_shuffle_epi32(A, 0);
 }
 
@@ -1128,13 +1128,13 @@ __m128d test_mm_shuffle_pd(__m128d A, __m128d B) {
 
 __m128i test_mm_shufflehi_epi16(__m128i A) {
   // CHECK-LABEL: test_mm_shufflehi_epi16
-  // CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> %{{.*}}, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 4, i32 4, i32 4>
+  // CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 4, i32 4, i32 4>
   return _mm_shufflehi_epi16(A, 0);
 }
 
 __m128i test_mm_shufflelo_epi16(__m128i A) {
   // CHECK-LABEL: test_mm_shufflelo_epi16
-  // CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> %{{.*}}, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7>
+  // CHECK: shufflevector <8 x i16> %{{.*}}, <8 x i16> undef, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 4, i32 5, i32 6, i32 7>
   return _mm_shufflelo_epi16(A, 0);
 }