[X86] Remove mask arguments from permvar builtins/intrinsics. Use a select in IR instead.

Someday maybe we'll use selects for all the builtins.

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@332825 91177308-0d34-0410-b5e6-96231b3b80d8

13 files changed, 163 insertions, 199 deletions

diff --git a/include/clang/Basic/BuiltinsX86.def b/include/clang/Basic/BuiltinsX86.def
index f7923919b3..56de0fd84e 100644
--- a/include/clang/Basic/BuiltinsX86.def
+++ b/include/clang/Basic/BuiltinsX86.def
@@ -1766,18 +1766,18 @@ TARGET_BUILTIN(__builtin_ia32_vfmsubsd3_mask3, "V2dV2dV2dV2dUcIi", "nc", "avx512
 TARGET_BUILTIN(__builtin_ia32_vfmsubss3_mask3, "V4fV4fV4fV4fUcIi", "nc", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_vfnmsubsd3_mask3, "V2dV2dV2dV2dUcIi", "nc", "avx512f")
 TARGET_BUILTIN(__builtin_ia32_vfnmsubss3_mask3, "V4fV4fV4fV4fUcIi", "nc", "avx512f")
-TARGET_BUILTIN(__builtin_ia32_permvarhi512_mask, "V32sV32sV32sV32sUi", "nc", "avx512bw")
-TARGET_BUILTIN(__builtin_ia32_permvardf512_mask, "V8dV8dV8LLiV8dUc", "nc", "avx512f")
-TARGET_BUILTIN(__builtin_ia32_permvardi512_mask, "V8LLiV8LLiV8LLiV8LLiUc", "nc", "avx512f")
-TARGET_BUILTIN(__builtin_ia32_permvarsf512_mask, "V16fV16fV16iV16fUs", "nc", "avx512f")
-TARGET_BUILTIN(__builtin_ia32_permvarsi512_mask, "V16iV16iV16iV16iUs", "nc", "avx512f")
-TARGET_BUILTIN(__builtin_ia32_permvarqi512_mask, "V64cV64cV64cV64cULLi", "nc", "avx512vbmi")
-TARGET_BUILTIN(__builtin_ia32_permvarqi128_mask, "V16cV16cV16cV16cUs", "nc", "avx512vbmi,avx512vl")
-TARGET_BUILTIN(__builtin_ia32_permvarqi256_mask, "V32cV32cV32cV32cUi", "nc", "avx512vbmi,avx512vl")
-TARGET_BUILTIN(__builtin_ia32_permvarhi128_mask, "V8sV8sV8sV8sUc", "nc", "avx512bw,avx512vl")
-TARGET_BUILTIN(__builtin_ia32_permvarhi256_mask, "V16sV16sV16sV16sUs", "nc", "avx512bw,avx512vl")
-TARGET_BUILTIN(__builtin_ia32_permvardf256_mask, "V4dV4dV4LLiV4dUc", "nc", "avx512vl")
-TARGET_BUILTIN(__builtin_ia32_permvardi256_mask, "V4LLiV4LLiV4LLiV4LLiUc", "nc", "avx512vl")
+TARGET_BUILTIN(__builtin_ia32_permvarhi512, "V32sV32sV32s", "nc", "avx512bw")
+TARGET_BUILTIN(__builtin_ia32_permvardf512, "V8dV8dV8LLi", "nc", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_permvardi512, "V8LLiV8LLiV8LLi", "nc", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_permvarsf512, "V16fV16fV16i", "nc", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_permvarsi512, "V16iV16iV16i", "nc", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_permvarqi512, "V64cV64cV64c", "nc", "avx512vbmi")
+TARGET_BUILTIN(__builtin_ia32_permvarqi128, "V16cV16cV16c", "nc", "avx512vbmi,avx512vl")
+TARGET_BUILTIN(__builtin_ia32_permvarqi256, "V32cV32cV32c", "nc", "avx512vbmi,avx512vl")
+TARGET_BUILTIN(__builtin_ia32_permvarhi128, "V8sV8sV8s", "nc", "avx512bw,avx512vl")
+TARGET_BUILTIN(__builtin_ia32_permvarhi256, "V16sV16sV16s", "nc", "avx512bw,avx512vl")
+TARGET_BUILTIN(__builtin_ia32_permvardf256, "V4dV4dV4LLi", "nc", "avx512vl")
+TARGET_BUILTIN(__builtin_ia32_permvardi256, "V4LLiV4LLiV4LLi", "nc", "avx512vl")
 TARGET_BUILTIN(__builtin_ia32_fpclasspd128_mask, "UcV2dIiUc", "nc", "avx512dq,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_fpclasspd256_mask, "UcV4dIiUc", "nc", "avx512dq,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_fpclassps128_mask, "UcV4fIiUc", "nc", "avx512dq,avx512vl")
diff --git a/lib/Headers/avx512bwintrin.h b/lib/Headers/avx512bwintrin.h
index 499bb8a690..fd1b871ae1 100644
--- a/lib/Headers/avx512bwintrin.h
+++ b/lib/Headers/avx512bwintrin.h
@@ -2034,30 +2034,25 @@ _mm512_maskz_broadcastw_epi16 (__mmask32 __M, __m128i __A)
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_permutexvar_epi16 (__m512i __A, __m512i __B)
 {
-  return (__m512i) __builtin_ia32_permvarhi512_mask ((__v32hi) __B,
-                 (__v32hi) __A,
-                 (__v32hi) _mm512_undefined_epi32 (),
-                 (__mmask32) -1);
+  return (__m512i)__builtin_ia32_permvarhi512((__v32hi)__B, (__v32hi)__A);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_maskz_permutexvar_epi16 (__mmask32 __M, __m512i __A,
         __m512i __B)
 {
-  return (__m512i) __builtin_ia32_permvarhi512_mask ((__v32hi) __B,
-                 (__v32hi) __A,
-                 (__v32hi) _mm512_setzero_hi(),
-                 (__mmask32) __M);
+  return (__m512i)__builtin_ia32_selectw_512((__mmask32)__M,
+                                    (__v32hi)_mm512_permutexvar_epi16(__A, __B),
+                                    (__v32hi)_mm512_setzero_hi());
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_permutexvar_epi16 (__m512i __W, __mmask32 __M, __m512i __A,
              __m512i __B)
 {
-  return (__m512i) __builtin_ia32_permvarhi512_mask ((__v32hi) __B,
-                 (__v32hi) __A,
-                 (__v32hi) __W,
-                 (__mmask32) __M);
+  return (__m512i)__builtin_ia32_selectw_512((__mmask32)__M,
+                                    (__v32hi)_mm512_permutexvar_epi16(__A, __B),
+                                    (__v32hi)__W);
 }
 
 #define _mm512_alignr_epi8(A, B, N) __extension__ ({\
diff --git a/lib/Headers/avx512fintrin.h b/lib/Headers/avx512fintrin.h
index ee5fb5ae0b..92d1ad26a9 100644
--- a/lib/Headers/avx512fintrin.h
+++ b/lib/Headers/avx512fintrin.h
@@ -8646,113 +8646,93 @@ _mm_mask3_fnmsub_sd (__m128d __W, __m128d __X, __m128d __Y, __mmask8 __U)
 static __inline__ __m512d __DEFAULT_FN_ATTRS
 _mm512_permutexvar_pd (__m512i __X, __m512d __Y)
 {
-  return (__m512d) __builtin_ia32_permvardf512_mask ((__v8df) __Y,
-                 (__v8di) __X,
-                 (__v8df) _mm512_undefined_pd (),
-                 (__mmask8) -1);
+  return (__m512d)__builtin_ia32_permvardf512((__v8df) __Y, (__v8di) __X);
 }
 
 static __inline__ __m512d __DEFAULT_FN_ATTRS
 _mm512_mask_permutexvar_pd (__m512d __W, __mmask8 __U, __m512i __X, __m512d __Y)
 {
-  return (__m512d) __builtin_ia32_permvardf512_mask ((__v8df) __Y,
-                 (__v8di) __X,
-                 (__v8df) __W,
-                 (__mmask8) __U);
+  return (__m512d)__builtin_ia32_selectpd_512((__mmask8)__U,
+                                        (__v8df)_mm512_permutexvar_pd(__X, __Y),
+                                        (__v8df)__W);
 }
 
 static __inline__ __m512d __DEFAULT_FN_ATTRS
 _mm512_maskz_permutexvar_pd (__mmask8 __U, __m512i __X, __m512d __Y)
 {
-  return (__m512d) __builtin_ia32_permvardf512_mask ((__v8df) __Y,
-                 (__v8di) __X,
-                 (__v8df) _mm512_setzero_pd (),
-                 (__mmask8) __U);
+  return (__m512d)__builtin_ia32_selectpd_512((__mmask8)__U,
+                                        (__v8df)_mm512_permutexvar_pd(__X, __Y),
+                                        (__v8df)_mm512_setzero_pd());
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
-_mm512_maskz_permutexvar_epi64 (__mmask8 __M, __m512i __X, __m512i __Y)
+_mm512_permutexvar_epi64 (__m512i __X, __m512i __Y)
 {
-  return (__m512i) __builtin_ia32_permvardi512_mask ((__v8di) __Y,
-                 (__v8di) __X,
-                 (__v8di) _mm512_setzero_si512 (),
-                 __M);
+  return (__m512i)__builtin_ia32_permvardi512((__v8di)__Y, (__v8di)__X);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
-_mm512_permutexvar_epi64 (__m512i __X, __m512i __Y)
+_mm512_maskz_permutexvar_epi64 (__mmask8 __M, __m512i __X, __m512i __Y)
 {
-  return (__m512i) __builtin_ia32_permvardi512_mask ((__v8di) __Y,
-                 (__v8di) __X,
-                 (__v8di) _mm512_undefined_epi32 (),
-                 (__mmask8) -1);
+  return (__m512i)__builtin_ia32_selectq_512((__mmask8)__M,
+                                     (__v8di)_mm512_permutexvar_epi64(__X, __Y),
+                                     (__v8di)_mm512_setzero_si512());
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_permutexvar_epi64 (__m512i __W, __mmask8 __M, __m512i __X,
              __m512i __Y)
 {
-  return (__m512i) __builtin_ia32_permvardi512_mask ((__v8di) __Y,
-                 (__v8di) __X,
-                 (__v8di) __W,
-                 __M);
+  return (__m512i)__builtin_ia32_selectq_512((__mmask8)__M,
+                                     (__v8di)_mm512_permutexvar_epi64(__X, __Y),
+                                     (__v8di)__W);
 }
 
 static __inline__ __m512 __DEFAULT_FN_ATTRS
 _mm512_permutexvar_ps (__m512i __X, __m512 __Y)
 {
-  return (__m512) __builtin_ia32_permvarsf512_mask ((__v16sf) __Y,
-                (__v16si) __X,
-                (__v16sf) _mm512_undefined_ps (),
-                (__mmask16) -1);
+  return (__m512)__builtin_ia32_permvarsf512((__v16sf)__Y, (__v16si)__X);
 }
 
 static __inline__ __m512 __DEFAULT_FN_ATTRS
 _mm512_mask_permutexvar_ps (__m512 __W, __mmask16 __U, __m512i __X, __m512 __Y)
 {
-  return (__m512) __builtin_ia32_permvarsf512_mask ((__v16sf) __Y,
-                (__v16si) __X,
-                (__v16sf) __W,
-                (__mmask16) __U);
+  return (__m512)__builtin_ia32_selectps_512((__mmask16)__U,
+                                       (__v16sf)_mm512_permutexvar_ps(__X, __Y),
+                                       (__v16sf)__W);
 }
 
 static __inline__ __m512 __DEFAULT_FN_ATTRS
 _mm512_maskz_permutexvar_ps (__mmask16 __U, __m512i __X, __m512 __Y)
 {
-  return (__m512) __builtin_ia32_permvarsf512_mask ((__v16sf) __Y,
-                (__v16si) __X,
-                (__v16sf) _mm512_setzero_ps (),
-                (__mmask16) __U);
+  return (__m512)__builtin_ia32_selectps_512((__mmask16)__U,
+                                       (__v16sf)_mm512_permutexvar_ps(__X, __Y),
+                                       (__v16sf)_mm512_setzero_ps());
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
-_mm512_maskz_permutexvar_epi32 (__mmask16 __M, __m512i __X, __m512i __Y)
+_mm512_permutexvar_epi32 (__m512i __X, __m512i __Y)
 {
-  return (__m512i) __builtin_ia32_permvarsi512_mask ((__v16si) __Y,
-                 (__v16si) __X,
-                 (__v16si) _mm512_setzero_si512 (),
-                 __M);
+  return (__m512i)__builtin_ia32_permvarsi512((__v16si)__Y, (__v16si)__X);
 }
 
+#define _mm512_permutevar_epi32 _mm512_permutexvar_epi32
+
 static __inline__ __m512i __DEFAULT_FN_ATTRS
-_mm512_permutexvar_epi32 (__m512i __X, __m512i __Y)
+_mm512_maskz_permutexvar_epi32 (__mmask16 __M, __m512i __X, __m512i __Y)
 {
-  return (__m512i) __builtin_ia32_permvarsi512_mask ((__v16si) __Y,
-                 (__v16si) __X,
-                 (__v16si) _mm512_undefined_epi32 (),
-                 (__mmask16) -1);
+  return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
+                                    (__v16si)_mm512_permutexvar_epi32(__X, __Y),
+                                    (__v16si)_mm512_setzero_si512());
 }
 
-#define _mm512_permutevar_epi32 _mm512_permutexvar_epi32
-
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_permutexvar_epi32 (__m512i __W, __mmask16 __M, __m512i __X,
              __m512i __Y)
 {
-  return (__m512i) __builtin_ia32_permvarsi512_mask ((__v16si) __Y,
-                 (__v16si) __X,
-                 (__v16si) __W,
-                 __M);
+  return (__m512i)__builtin_ia32_selectd_512((__mmask16)__M,
+                                    (__v16si)_mm512_permutexvar_epi32(__X, __Y),
+                                    (__v16si)__W);
 }
 
 #define _mm512_mask_permutevar_epi32 _mm512_mask_permutexvar_epi32
diff --git a/lib/Headers/avx512vbmiintrin.h b/lib/Headers/avx512vbmiintrin.h
index 837238eda9..a8ae95c7f2 100644
--- a/lib/Headers/avx512vbmiintrin.h
+++ b/lib/Headers/avx512vbmiintrin.h
@@ -78,30 +78,25 @@ _mm512_maskz_permutex2var_epi8 (__mmask64 __U, __m512i __A,
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_permutexvar_epi8 (__m512i __A, __m512i __B)
 {
-  return (__m512i) __builtin_ia32_permvarqi512_mask ((__v64qi) __B,
-                 (__v64qi) __A,
-                 (__v64qi) _mm512_undefined_epi32 (),
-                 (__mmask64) -1);
+  return (__m512i)__builtin_ia32_permvarqi512((__v64qi) __B, (__v64qi) __A);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_maskz_permutexvar_epi8 (__mmask64 __M, __m512i __A,
         __m512i __B)
 {
-  return (__m512i) __builtin_ia32_permvarqi512_mask ((__v64qi) __B,
-                 (__v64qi) __A,
-                 (__v64qi) _mm512_setzero_si512(),
-                 (__mmask64) __M);
+  return (__m512i)__builtin_ia32_selectb_512((__mmask64)__M,
+                                     (__v64qi)_mm512_permutexvar_epi8(__A, __B),
+                                     (__v64qi)_mm512_setzero_si512());
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
 _mm512_mask_permutexvar_epi8 (__m512i __W, __mmask64 __M, __m512i __A,
              __m512i __B)
 {
-  return (__m512i) __builtin_ia32_permvarqi512_mask ((__v64qi) __B,
-                 (__v64qi) __A,
-                 (__v64qi) __W,
-                 (__mmask64) __M);
+  return (__m512i)__builtin_ia32_selectb_512((__mmask64)__M,
+                                     (__v64qi)_mm512_permutexvar_epi8(__A, __B),
+                                     (__v64qi)__W);
 }
 
 static __inline__ __m512i __DEFAULT_FN_ATTRS
diff --git a/lib/Headers/avx512vbmivlintrin.h b/lib/Headers/avx512vbmivlintrin.h
index 105c6d142f..b2336554ef 100644
--- a/lib/Headers/avx512vbmivlintrin.h
+++ b/lib/Headers/avx512vbmivlintrin.h
@@ -129,58 +129,48 @@ _mm256_maskz_permutex2var_epi8 (__mmask32 __U, __m256i __A,
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_permutexvar_epi8 (__m128i __A, __m128i __B)
 {
-  return (__m128i) __builtin_ia32_permvarqi128_mask ((__v16qi) __B,
-                 (__v16qi) __A,
-                 (__v16qi) _mm_undefined_si128 (),
-                 (__mmask16) -1);
+  return (__m128i)__builtin_ia32_permvarqi128((__v16qi)__B, (__v16qi)__A);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_maskz_permutexvar_epi8 (__mmask16 __M, __m128i __A, __m128i __B)
 {
-  return (__m128i) __builtin_ia32_permvarqi128_mask ((__v16qi) __B,
-                 (__v16qi) __A,
-                 (__v16qi) _mm_setzero_si128 (),
-                 (__mmask16) __M);
+  return (__m128i)__builtin_ia32_selectb_128((__mmask16)__M,
+                                        (__v16qi)_mm_permutexvar_epi8(__A, __B),
+                                        (__v16qi)_mm_setzero_si128());
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_permutexvar_epi8 (__m128i __W, __mmask16 __M, __m128i __A,
           __m128i __B)
 {
-  return (__m128i) __builtin_ia32_permvarqi128_mask ((__v16qi) __B,
-                 (__v16qi) __A,
-                 (__v16qi) __W,
-                 (__mmask16) __M);
+  return (__m128i)__builtin_ia32_selectb_128((__mmask16)__M,
+                                        (__v16qi)_mm_permutexvar_epi8(__A, __B),
+                                        (__v16qi)__W);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_permutexvar_epi8 (__m256i __A, __m256i __B)
 {
-  return (__m256i) __builtin_ia32_permvarqi256_mask ((__v32qi) __B,
-                 (__v32qi) __A,
-                 (__v32qi) _mm256_undefined_si256 (),
-                 (__mmask32) -1);
+  return (__m256i)__builtin_ia32_permvarqi256((__v32qi) __B, (__v32qi) __A);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_maskz_permutexvar_epi8 (__mmask32 __M, __m256i __A,
         __m256i __B)
 {
-  return (__m256i) __builtin_ia32_permvarqi256_mask ((__v32qi) __B,
-                 (__v32qi) __A,
-                 (__v32qi) _mm256_setzero_si256 (),
-                 (__mmask32) __M);
+  return (__m256i)__builtin_ia32_selectb_256((__mmask32)__M,
+                                     (__v32qi)_mm256_permutexvar_epi8(__A, __B),
+                                     (__v32qi)_mm256_setzero_si256());
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mask_permutexvar_epi8 (__m256i __W, __mmask32 __M, __m256i __A,
              __m256i __B)
 {
-  return (__m256i) __builtin_ia32_permvarqi256_mask ((__v32qi) __B,
-                 (__v32qi) __A,
-                 (__v32qi) __W,
-                 (__mmask32) __M);
+  return (__m256i)__builtin_ia32_selectb_256((__mmask32)__M,
+                                     (__v32qi)_mm256_permutexvar_epi8(__A, __B),
+                                     (__v32qi)__W);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
diff --git a/lib/Headers/avx512vlbwintrin.h b/lib/Headers/avx512vlbwintrin.h
index 36b19b99ab..6c2d3eadc9 100644
--- a/lib/Headers/avx512vlbwintrin.h
+++ b/lib/Headers/avx512vlbwintrin.h
@@ -2664,58 +2664,48 @@ _mm_maskz_set1_epi16 (__mmask8 __M, short __A)
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_permutexvar_epi16 (__m128i __A, __m128i __B)
 {
-  return (__m128i) __builtin_ia32_permvarhi128_mask ((__v8hi) __B,
-                 (__v8hi) __A,
-                 (__v8hi) _mm_undefined_si128 (),
-                 (__mmask8) -1);
+  return (__m128i)__builtin_ia32_permvarhi128((__v8hi) __B, (__v8hi) __A);
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_maskz_permutexvar_epi16 (__mmask8 __M, __m128i __A, __m128i __B)
 {
-  return (__m128i) __builtin_ia32_permvarhi128_mask ((__v8hi) __B,
-                 (__v8hi) __A,
-                 (__v8hi) _mm_setzero_si128 (),
-                 (__mmask8) __M);
+  return (__m128i)__builtin_ia32_selectw_128((__mmask8)__M,
+                                        (__v8hi)_mm_permutexvar_epi16(__A, __B),
+                                        (__v8hi) _mm_setzero_si128());
 }
 
 static __inline__ __m128i __DEFAULT_FN_ATTRS
 _mm_mask_permutexvar_epi16 (__m128i __W, __mmask8 __M, __m128i __A,
           __m128i __B)
 {
-  return (__m128i) __builtin_ia32_permvarhi128_mask ((__v8hi) __B,
-                 (__v8hi) __A,
-                 (__v8hi) __W,
-                 (__mmask8) __M);
+  return (__m128i)__builtin_ia32_selectw_128((__mmask8)__M,
+                                        (__v8hi)_mm_permutexvar_epi16(__A, __B),
+                                        (__v8hi)__W);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_permutexvar_epi16 (__m256i __A, __m256i __B)
 {
-  return (__m256i) __builtin_ia32_permvarhi256_mask ((__v16hi) __B,
-                 (__v16hi) __A,
-                 (__v16hi) _mm256_undefined_si256 (),
-                 (__mmask16) -1);
+  return (__m256i)__builtin_ia32_permvarhi256((__v16hi) __B, (__v16hi) __A);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_maskz_permutexvar_epi16 (__mmask16 __M, __m256i __A,
         __m256i __B)
 {
-  return (__m256i) __builtin_ia32_permvarhi256_mask ((__v16hi) __B,
-                 (__v16hi) __A,
-                 (__v16hi) _mm256_setzero_si256 (),
-                 (__mmask16) __M);
+  return (__m256i)__builtin_ia32_selectw_256((__mmask16)__M,
+                                    (__v16hi)_mm256_permutexvar_epi16(__A, __B),
+                                    (__v16hi)_mm256_setzero_si256());
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mask_permutexvar_epi16 (__m256i __W, __mmask16 __M, __m256i __A,
              __m256i __B)
 {
-  return (__m256i) __builtin_ia32_permvarhi256_mask ((__v16hi) __B,
-                 (__v16hi) __A,
-                 (__v16hi) __W,
-                 (__mmask16) __M);
+  return (__m256i)__builtin_ia32_selectw_256((__mmask16)__M,
+                                    (__v16hi)_mm256_permutexvar_epi16(__A, __B),
+                                    (__v16hi)__W);
 }
 
 #define _mm_mask_alignr_epi8(W, U, A, B, N) __extension__ ({ \
diff --git a/lib/Headers/avx512vlintrin.h b/lib/Headers/avx512vlintrin.h
index de89cfdfac..213d911fd5 100644
--- a/lib/Headers/avx512vlintrin.h
+++ b/lib/Headers/avx512vlintrin.h
@@ -8108,57 +8108,47 @@ _mm256_mask_cvtepi64_storeu_epi16 (void * __P, __mmask8 __M, __m256i __A)
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_permutexvar_pd (__m256i __X, __m256d __Y)
 {
-  return (__m256d) __builtin_ia32_permvardf256_mask ((__v4df) __Y,
-                 (__v4di) __X,
-                 (__v4df) _mm256_undefined_si256 (),
-                 (__mmask8) -1);
+  return (__m256d)__builtin_ia32_permvardf256((__v4df)__Y, (__v4di)__X);
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_mask_permutexvar_pd (__m256d __W, __mmask8 __U, __m256i __X,
           __m256d __Y)
 {
-  return (__m256d) __builtin_ia32_permvardf256_mask ((__v4df) __Y,
-                 (__v4di) __X,
-                 (__v4df) __W,
-                 (__mmask8) __U);
+  return (__m256d)__builtin_ia32_selectpd_256((__mmask8)__U,
+                                        (__v4df)_mm256_permutexvar_pd(__X, __Y),
+                                        (__v4df)__W);
 }
 
 static __inline__ __m256d __DEFAULT_FN_ATTRS
 _mm256_maskz_permutexvar_pd (__mmask8 __U, __m256i __X, __m256d __Y)
 {
-  return (__m256d) __builtin_ia32_permvardf256_mask ((__v4df) __Y,
-                 (__v4di) __X,
-                 (__v4df) _mm256_setzero_pd (),
-                 (__mmask8) __U);
+  return (__m256d)__builtin_ia32_selectpd_256((__mmask8)__U,
+                                        (__v4df)_mm256_permutexvar_pd(__X, __Y),
+                                        (__v4df)_mm256_setzero_pd());
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
-_mm256_maskz_permutexvar_epi64 (__mmask8 __M, __m256i __X, __m256i __Y)
+_mm256_permutexvar_epi64 ( __m256i __X, __m256i __Y)
 {
-  return (__m256i) __builtin_ia32_permvardi256_mask ((__v4di) __Y,
-                 (__v4di) __X,
-                 (__v4di) _mm256_setzero_si256 (),
-                 (__mmask8) __M);
+  return (__m256i)__builtin_ia32_permvardi256((__v4di) __Y, (__v4di) __X);
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
-_mm256_permutexvar_epi64 ( __m256i __X, __m256i __Y)
+_mm256_maskz_permutexvar_epi64 (__mmask8 __M, __m256i __X, __m256i __Y)
 {
-  return (__m256i) __builtin_ia32_permvardi256_mask ((__v4di) __Y,
-                 (__v4di) __X,
-                 (__v4di) _mm256_undefined_si256 (),
-                 (__mmask8) -1);
+  return (__m256i)__builtin_ia32_selectq_256((__mmask8)__M,
+                                     (__v4di)_mm256_permutexvar_epi64(__X, __Y),
+                                     (__v4di)_mm256_setzero_si256());
 }
 
 static __inline__ __m256i __DEFAULT_FN_ATTRS
 _mm256_mask_permutexvar_epi64 (__m256i __W, __mmask8 __M, __m256i __X,
              __m256i __Y)
 {
-  return (__m256i) __builtin_ia32_permvardi256_mask ((__v4di) __Y,
-                 (__v4di) __X,
-                 (__v4di) __W,
-                 __M);
+  return (__m256i)__builtin_ia32_selectq_256((__mmask8)__M,
+                                     (__v4di)_mm256_permutexvar_epi64(__X, __Y),
+                                     (__v4di)__W);
 }
 
 #define _mm256_permutexvar_ps(A, B) _mm256_permutevar8x32_ps((B), (A))
diff --git a/test/CodeGen/avx512bw-builtins.c b/test/CodeGen/avx512bw-builtins.c
index ee2ae6ad29..d5c1993998 100644
--- a/test/CodeGen/avx512bw-builtins.c
+++ b/test/CodeGen/avx512bw-builtins.c
@@ -1890,19 +1890,21 @@ __m512i test_mm512_maskz_set1_epi16(__mmask32 __M, short __A) {
 }
 __m512i test_mm512_permutexvar_epi16(__m512i __A, __m512i __B) {
   // CHECK-LABEL: @test_mm512_permutexvar_epi16
-  // CHECK: @llvm.x86.avx512.mask.permvar.hi.512
+  // CHECK: @llvm.x86.avx512.permvar.hi.512
  return _mm512_permutexvar_epi16(__A, __B); 
 }
 
 __m512i test_mm512_maskz_permutexvar_epi16(__mmask32 __M, __m512i __A, __m512i __B) {
  // CHECK-LABEL: @test_mm512_maskz_permutexvar_epi16
-  // CHECK: @llvm.x86.avx512.mask.permvar.hi.512
+  // CHECK: @llvm.x86.avx512.permvar.hi.512
+  // CHECK: select <32 x i1> %{{.*}}, <32 x i16> %{{.*}}, <32 x i16> %{{.*}}
   return _mm512_maskz_permutexvar_epi16(__M, __A, __B); 
 }
 
 __m512i test_mm512_mask_permutexvar_epi16(__m512i __W, __mmask32 __M, __m512i __A, __m512i __B) {
   // CHECK-LABEL: @test_mm512_mask_permutexvar_epi16
-  // CHECK: @llvm.x86.avx512.mask.permvar.hi.512
+  // CHECK: @llvm.x86.avx512.permvar.hi.512
+  // CHECK: select <32 x i1> %{{.*}}, <32 x i16> %{{.*}}, <32 x i16> %{{.*}}
   return _mm512_mask_permutexvar_epi16(__W, __M, __A, __B); 
 }
 __m512i test_mm512_alignr_epi8(__m512i __A,__m512i __B){
diff --git a/test/CodeGen/avx512f-builtins.c b/test/CodeGen/avx512f-builtins.c
index 0bff5c8dbd..a52adfd269 100644
--- a/test/CodeGen/avx512f-builtins.c
+++ b/test/CodeGen/avx512f-builtins.c
@@ -6180,73 +6180,81 @@ __m512i test_mm512_maskz_permutex_epi64(__mmask8 __M, __m512i __X) {
 
 __m512d test_mm512_permutexvar_pd(__m512i __X, __m512d __Y) {
   // CHECK-LABEL: @test_mm512_permutexvar_pd
-  // CHECK: @llvm.x86.avx512.mask.permvar.df.512
+  // CHECK: @llvm.x86.avx512.permvar.df.512
   return _mm512_permutexvar_pd(__X, __Y); 
 }
 
 __m512d test_mm512_mask_permutexvar_pd(__m512d __W, __mmask8 __U, __m512i __X, __m512d __Y) {
   // CHECK-LABEL: @test_mm512_mask_permutexvar_pd
-  // CHECK: @llvm.x86.avx512.mask.permvar.df.512
+  // CHECK: @llvm.x86.avx512.permvar.df.512
+  // CHECK: select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
   return _mm512_mask_permutexvar_pd(__W, __U, __X, __Y); 
 }
 
 __m512d test_mm512_maskz_permutexvar_pd(__mmask8 __U, __m512i __X, __m512d __Y) {
   // CHECK-LABEL: @test_mm512_maskz_permutexvar_pd
-  // CHECK: @llvm.x86.avx512.mask.permvar.df.512
+  // CHECK: @llvm.x86.avx512.permvar.df.512
+  // CHECK: select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
   return _mm512_maskz_permutexvar_pd(__U, __X, __Y); 
 }
 
 __m512i test_mm512_maskz_permutexvar_epi64(__mmask8 __M, __m512i __X, __m512i __Y) {
   // CHECK-LABEL: @test_mm512_maskz_permutexvar_epi64
-  // CHECK: @llvm.x86.avx512.mask.permvar.di.512
+  // CHECK: @llvm.x86.avx512.permvar.di.512
+  // CHECK: select <8 x i1> %{{.*}}, <8 x i64> %{{.*}}, <8 x i64> %{{.*}}
   return _mm512_maskz_permutexvar_epi64(__M, __X, __Y); 
 }
 
 __m512i test_mm512_permutexvar_epi64(__m512i __X, __m512i __Y) {
   // CHECK-LABEL: @test_mm512_permutexvar_epi64
-  // CHECK: @llvm.x86.avx512.mask.permvar.di.512
+  // CHECK: @llvm.x86.avx512.permvar.di.512
   return _mm512_permutexvar_epi64(__X, __Y); 
 }
 
 __m512i test_mm512_mask_permutexvar_epi64(__m512i __W, __mmask8 __M, __m512i __X, __m512i __Y) {
   // CHECK-LABEL: @test_mm512_mask_permutexvar_epi64
-  // CHECK: @llvm.x86.avx512.mask.permvar.di.512
+  // CHECK: @llvm.x86.avx512.permvar.di.512
+  // CHECK: select <8 x i1> %{{.*}}, <8 x i64> %{{.*}}, <8 x i64> %{{.*}}
   return _mm512_mask_permutexvar_epi64(__W, __M, __X, __Y); 
 }
 
 __m512 test_mm512_permutexvar_ps(__m512i __X, __m512 __Y) {
   // CHECK-LABEL: @test_mm512_permutexvar_ps
-  // CHECK: @llvm.x86.avx512.mask.permvar.sf.512
+  // CHECK: @llvm.x86.avx512.permvar.sf.512
   return _mm512_permutexvar_ps(__X, __Y); 
 }
 
 __m512 test_mm512_mask_permutexvar_ps(__m512 __W, __mmask16 __U, __m512i __X, __m512 __Y) {
   // CHECK-LABEL: @test_mm512_mask_permutexvar_ps
-  // CHECK: @llvm.x86.avx512.mask.permvar.sf.512
+  // CHECK: @llvm.x86.avx512.permvar.sf.512
+  // CHECK: select <16 x i1> %{{.*}}, <16 x float> %{{.*}}, <16 x float> %{{.*}}
   return _mm512_mask_permutexvar_ps(__W, __U, __X, __Y); 
 }
 
 __m512 test_mm512_maskz_permutexvar_ps(__mmask16 __U, __m512i __X, __m512 __Y) {
   // CHECK-LABEL: @test_mm512_maskz_permutexvar_ps
-  // CHECK: @llvm.x86.avx512.mask.permvar.sf.512
+  // CHECK: @llvm.x86.avx512.permvar.sf.512
+  // CHECK: select <16 x i1> %{{.*}}, <16 x float> %{{.*}}, <16 x float> %{{.*}}
   return _mm512_maskz_permutexvar_ps(__U, __X, __Y); 
 }
 
 __m512i test_mm512_maskz_permutexvar_epi32(__mmask16 __M, __m512i __X, __m512i __Y) {
   // CHECK-LABEL: @test_mm512_maskz_permutexvar_epi32
-  // CHECK: @llvm.x86.avx512.mask.permvar.si.512
+  // CHECK: @llvm.x86.avx512.permvar.si.512
+  // CHECK: select <16 x i1> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> %{{.*}}
   return _mm512_maskz_permutexvar_epi32(__M, __X, __Y); 
 }
 
 __m512i test_mm512_permutexvar_epi32(__m512i __X, __m512i __Y) {
   // CHECK-LABEL: @test_mm512_permutexvar_epi32
-  // CHECK: @llvm.x86.avx512.mask.permvar.si.512
+  // CHECK: @llvm.x86.avx512.permvar.si.512
   return _mm512_permutexvar_epi32(__X, __Y); 
 }
 
 __m512i test_mm512_mask_permutexvar_epi32(__m512i __W, __mmask16 __M, __m512i __X, __m512i __Y) {
   // CHECK-LABEL: @test_mm512_mask_permutexvar_epi32
-  // CHECK: @llvm.x86.avx512.mask.permvar.si.512
+  // CHECK: @llvm.x86.avx512.permvar.si.512
+  // CHECK: select <16 x i1> %{{.*}}, <16 x i32> %{{.*}}, <16 x i32> %{{.*}}
   return _mm512_mask_permutexvar_epi32(__W, __M, __X, __Y); 
 }
 
diff --git a/test/CodeGen/avx512vbmi-builtins.c b/test/CodeGen/avx512vbmi-builtins.c
index 0816bce3a6..d670f3f8ab 100644
--- a/test/CodeGen/avx512vbmi-builtins.c
+++ b/test/CodeGen/avx512vbmi-builtins.c
@@ -29,19 +29,21 @@ __m512i test_mm512_maskz_permutex2var_epi8(__mmask64 __U, __m512i __A, __m512i _
 
 __m512i test_mm512_permutexvar_epi8(__m512i __A, __m512i __B) {
   // CHECK-LABEL: @test_mm512_permutexvar_epi8
-  // CHECK: @llvm.x86.avx512.mask.permvar.qi.512
+  // CHECK: @llvm.x86.avx512.permvar.qi.512
   return _mm512_permutexvar_epi8(__A, __B); 
 }
 
 __m512i test_mm512_maskz_permutexvar_epi8(__mmask64 __M, __m512i __A, __m512i __B) {
   // CHECK-LABEL: @test_mm512_maskz_permutexvar_epi8
-  // CHECK: @llvm.x86.avx512.mask.permvar.qi.512
+  // CHECK: @llvm.x86.avx512.permvar.qi.512
+  // CHECK: select <64 x i1> %{{.*}}, <64 x i8> %{{.*}}, <64 x i8> %{{.*}}
   return _mm512_maskz_permutexvar_epi8(__M, __A, __B); 
 }
 
 __m512i test_mm512_mask_permutexvar_epi8(__m512i __W, __mmask64 __M, __m512i __A, __m512i __B) {
   // CHECK-LABEL: @test_mm512_mask_permutexvar_epi8
-  // CHECK: @llvm.x86.avx512.mask.permvar.qi.512
+  // CHECK: @llvm.x86.avx512.permvar.qi.512
+  // CHECK: select <64 x i1> %{{.*}}, <64 x i8> %{{.*}}, <64 x i8> %{{.*}}
   return _mm512_mask_permutexvar_epi8(__W, __M, __A, __B); 
 }
 
diff --git a/test/CodeGen/avx512vbmivl-builtin.c b/test/CodeGen/avx512vbmivl-builtin.c
index b114720758..05aa626fbd 100644
--- a/test/CodeGen/avx512vbmivl-builtin.c
+++ b/test/CodeGen/avx512vbmivl-builtin.c
@@ -5,37 +5,41 @@
 
 __m128i test_mm_permutexvar_epi8(__m128i __A, __m128i __B) {
   // CHECK-LABEL: @test_mm_permutexvar_epi8
-  // CHECK: @llvm.x86.avx512.mask.permvar.qi.128
+  // CHECK: @llvm.x86.avx512.permvar.qi.128
   return _mm_permutexvar_epi8(__A, __B); 
 }
 
 __m128i test_mm_maskz_permutexvar_epi8(__mmask16 __M, __m128i __A, __m128i __B) {
   // CHECK-LABEL: @test_mm_maskz_permutexvar_epi8
-  // CHECK: @llvm.x86.avx512.mask.permvar.qi.128
+  // CHECK: @llvm.x86.avx512.permvar.qi.128
+  // CHECK: select <16 x i1> %{{.*}}, <16 x i8> %{{.*}}, <16 x i8> %{{.*}}
   return _mm_maskz_permutexvar_epi8(__M, __A, __B); 
 }
 
 __m128i test_mm_mask_permutexvar_epi8(__m128i __W, __mmask16 __M, __m128i __A, __m128i __B) {
   // CHECK-LABEL: @test_mm_mask_permutexvar_epi8
-  // CHECK: @llvm.x86.avx512.mask.permvar.qi.128
+  // CHECK: @llvm.x86.avx512.permvar.qi.128
+  // CHECK: select <16 x i1> %{{.*}}, <16 x i8> %{{.*}}, <16 x i8> %{{.*}}
   return _mm_mask_permutexvar_epi8(__W, __M, __A, __B); 
 }
 
 __m256i test_mm256_permutexvar_epi8(__m256i __A, __m256i __B) {
   // CHECK-LABEL: @test_mm256_permutexvar_epi8
-  // CHECK: @llvm.x86.avx512.mask.permvar.qi.256
+  // CHECK: @llvm.x86.avx512.permvar.qi.256
   return _mm256_permutexvar_epi8(__A, __B); 
 }
 
 __m256i test_mm256_maskz_permutexvar_epi8(__mmask32 __M, __m256i __A, __m256i __B) {
   // CHECK-LABEL: @test_mm256_maskz_permutexvar_epi8
-  // CHECK: @llvm.x86.avx512.mask.permvar.qi.256
+  // CHECK: @llvm.x86.avx512.permvar.qi.256
+  // CHECK: select <32 x i1> %{{.*}}, <32 x i8> %{{.*}}, <32 x i8> %{{.*}}
   return _mm256_maskz_permutexvar_epi8(__M, __A, __B); 
 }
 
 __m256i test_mm256_mask_permutexvar_epi8(__m256i __W, __mmask32 __M, __m256i __A, __m256i __B) {
   // CHECK-LABEL: @test_mm256_mask_permutexvar_epi8
-  // CHECK: @llvm.x86.avx512.mask.permvar.qi.256
+  // CHECK: @llvm.x86.avx512.permvar.qi.256
+  // CHECK: select <32 x i1> %{{.*}}, <32 x i8> %{{.*}}, <32 x i8> %{{.*}}
   return _mm256_mask_permutexvar_epi8(__W, __M, __A, __B); 
 }
 
diff --git a/test/CodeGen/avx512vl-builtins.c b/test/CodeGen/avx512vl-builtins.c
index 27c0993b9b..7771a49c9b 100644
--- a/test/CodeGen/avx512vl-builtins.c
+++ b/test/CodeGen/avx512vl-builtins.c
@@ -7035,31 +7035,35 @@ __m256i test_mm256_maskz_permutex_epi64(__mmask8 __M, __m256i __X) {
 
 __m256d test_mm256_permutexvar_pd(__m256i __X, __m256d __Y) {
   // CHECK-LABEL: @test_mm256_permutexvar_pd
-  // CHECK: @llvm.x86.avx512.mask.permvar.df.256
+  // CHECK: @llvm.x86.avx512.permvar.df.256
   return _mm256_permutexvar_pd(__X, __Y);
 }
 
 __m256d test_mm256_mask_permutexvar_pd(__m256d __W, __mmask8 __U, __m256i __X, __m256d __Y) {
   // CHECK-LABEL: @test_mm256_mask_permutexvar_pd
-  // CHECK: @llvm.x86.avx512.mask.permvar.df.256
+  // CHECK: @llvm.x86.avx512.permvar.df.256
+  // CHECK: select <4 x i1> %{{.*}}, <4 x double> %{{.*}}, <4 x double> %{{.*}}
   return _mm256_mask_permutexvar_pd(__W, __U, __X, __Y);
 }
 
 __m256d test_mm256_maskz_permutexvar_pd(__mmask8 __U, __m256i __X, __m256d __Y) {
   // CHECK-LABEL: @test_mm256_maskz_permutexvar_pd
-  // CHECK: @llvm.x86.avx512.mask.permvar.df.256
+  // CHECK: @llvm.x86.avx512.permvar.df.256
+  // CHECK: select <4 x i1> %{{.*}}, <4 x double> %{{.*}}, <4 x double> %{{.*}}
   return _mm256_maskz_permutexvar_pd(__U, __X, __Y);
 }
 
 __m256i test_mm256_maskz_permutexvar_epi64(__mmask8 __M, __m256i __X, __m256i __Y) {
   // CHECK-LABEL: @test_mm256_maskz_permutexvar_epi64
-  // CHECK: @llvm.x86.avx512.mask.permvar.di.256
+  // CHECK: @llvm.x86.avx512.permvar.di.256
+  // CHECK: select <4 x i1> %{{.*}}, <4 x i64> %{{.*}}, <4 x i64> %{{.*}}
   return _mm256_maskz_permutexvar_epi64(__M, __X, __Y);
 }
 
 __m256i test_mm256_mask_permutexvar_epi64(__m256i __W, __mmask8 __M, __m256i __X, __m256i __Y) {
   // CHECK-LABEL: @test_mm256_mask_permutexvar_epi64
-  // CHECK: @llvm.x86.avx512.mask.permvar.di.256
+  // CHECK: @llvm.x86.avx512.permvar.di.256
+  // CHECK: select <4 x i1> %{{.*}}, <4 x i64> %{{.*}}, <4 x i64> %{{.*}}
   return _mm256_mask_permutexvar_epi64(__W, __M, __X, __Y);
 }
 
diff --git a/test/CodeGen/avx512vlbw-builtins.c b/test/CodeGen/avx512vlbw-builtins.c
index 1118b958e2..8ca5fc9555 100644
--- a/test/CodeGen/avx512vlbw-builtins.c
+++ b/test/CodeGen/avx512vlbw-builtins.c
@@ -2891,37 +2891,41 @@ __m128i test_mm_maskz_set1_epi16(__mmask8 __M, short __A) {
 }
 __m128i test_mm_permutexvar_epi16(__m128i __A, __m128i __B) {
   // CHECK-LABEL: @test_mm_permutexvar_epi16
-  // CHECK: @llvm.x86.avx512.mask.permvar.hi.128
+  // CHECK: @llvm.x86.avx512.permvar.hi.128
   return _mm_permutexvar_epi16(__A, __B); 
 }
 
 __m128i test_mm_maskz_permutexvar_epi16(__mmask8 __M, __m128i __A, __m128i __B) {
   // CHECK-LABEL: @test_mm_maskz_permutexvar_epi16
-  // CHECK: @llvm.x86.avx512.mask.permvar.hi.128
+  // CHECK: @llvm.x86.avx512.permvar.hi.128
+  // CHECK: select <8 x i1> %{{.*}}, <8 x i16> %{{.*}}, <8 x i16> %{{.*}}
   return _mm_maskz_permutexvar_epi16(__M, __A, __B); 
 }
 
 __m128i test_mm_mask_permutexvar_epi16(__m128i __W, __mmask8 __M, __m128i __A, __m128i __B) {
   // CHECK-LABEL: @test_mm_mask_permutexvar_epi16
-  // CHECK: @llvm.x86.avx512.mask.permvar.hi.128
+  // CHECK: @llvm.x86.avx512.permvar.hi.128
+  // CHECK: select <8 x i1> %{{.*}}, <8 x i16> %{{.*}}, <8 x i16> %{{.*}}
   return _mm_mask_permutexvar_epi16(__W, __M, __A, __B); 
 }
 
 __m256i test_mm256_permutexvar_epi16(__m256i __A, __m256i __B) {
   // CHECK-LABEL: @test_mm256_permutexvar_epi16
-  // CHECK: @llvm.x86.avx512.mask.permvar.hi.256
+  // CHECK: @llvm.x86.avx512.permvar.hi.256
   return _mm256_permutexvar_epi16(__A, __B); 
 }
 
 __m256i test_mm256_maskz_permutexvar_epi16(__mmask16 __M, __m256i __A, __m256i __B) {
   // CHECK-LABEL: @test_mm256_maskz_permutexvar_epi16
-  // CHECK: @llvm.x86.avx512.mask.permvar.hi.256
+  // CHECK: @llvm.x86.avx512.permvar.hi.256
+  // CHECK: select <16 x i1> %{{.*}}, <16 x i16> %{{.*}}, <16 x i16> %{{.*}}
   return _mm256_maskz_permutexvar_epi16(__M, __A, __B); 
 }
 
 __m256i test_mm256_mask_permutexvar_epi16(__m256i __W, __mmask16 __M, __m256i __A, __m256i __B) {
   // CHECK-LABEL: @test_mm256_mask_permutexvar_epi16
-  // CHECK: @llvm.x86.avx512.mask.permvar.hi.256
+  // CHECK: @llvm.x86.avx512.permvar.hi.256
+  // CHECK: select <16 x i1> %{{.*}}, <16 x i16> %{{.*}}, <16 x i16> %{{.*}}
   return _mm256_mask_permutexvar_epi16(__W, __M, __A, __B); 
 }
 __m128i test_mm_mask_alignr_epi8(__m128i __W, __mmask16 __U, __m128i __A, __m128i __B) {