[CLANG][AVX512][BUILTIN]movap{d|s}{128|256|512}

[delena]

include/clang/Basic/BuiltinsX86.def

9

it should be under VL

lib/Headers/avx512fintrin.h

36

please remove 2 empty lines

lib/Headers/avx512vlintrin.h

72

and here

include/clang/Basic/BuiltinsX86.def

 //===--- BuiltinsX86.def - X86 Builtin function database --------*- C++ -*-===//
 //
 //                     The LLVM Compiler Infrastructure
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 //
 // This file defines the X86-specific builtin function database.  Users of
 // this file must define the BUILTIN macro to make use of this information.
 //
 //===----------------------------------------------------------------------===//
 
 // The format of this database matches clang/Basic/Builtins.def.
 
 // FIXME: Ideally we would be able to pull this information from what

 TARGET_BUILTIN(__builtin_ia32_psravq128_mask, "V2LLiV2LLiV2LLiV2LLiUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_psravq256_mask, "V4LLiV4LLiV4LLiV4LLiUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_psraw512_mask, "V32sV32sV8sV32sUi","","avx512bw")
 TARGET_BUILTIN(__builtin_ia32_psrawi512_mask, "V32sV32sIiV32sUi","","avx512bw")
 TARGET_BUILTIN(__builtin_ia32_psraw128_mask, "V8sV8sV8sV8sUc","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_psraw256_mask, "V16sV16sV8sV16sUs","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_psrawi128_mask, "V8sV8sIiV8sUc","","avx512bw,avx512vl")
 TARGET_BUILTIN(__builtin_ia32_psrawi256_mask, "V16sV16sIiV16sUs","","avx512bw,avx512vl")
+TARGET_BUILTIN(__builtin_ia32_movapd512_mask, "V8dV8dV8dUc","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_movaps512_mask, "V16fV16fV16fUs","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_movapd128_mask, "V2dV2dV2dUc","","avx512vl")
+TARGET_BUILTIN(__builtin_ia32_movapd256_mask, "V4dV4dV4dUc","","avx512vl")
+TARGET_BUILTIN(__builtin_ia32_movaps128_mask, "V4fV4fV4fUc","","avx512vl")
+TARGET_BUILTIN(__builtin_ia32_movaps256_mask, "V8fV8fV8fUc","","avx512vl")
 
 #undef BUILTIN
 #undef TARGET_BUILTIN

lib/Headers/avx512fintrin.h

 #define __AVX512FINTRIN_H
 
 typedef double __v8df __attribute__((__vector_size__(64)));
 typedef float __v16sf __attribute__((__vector_size__(64)));
 typedef long long __v8di __attribute__((__vector_size__(64)));
 typedef int __v16si __attribute__((__vector_size__(64)));
 
 typedef float __m512 __attribute__((__vector_size__(64)));
 typedef double __m512d __attribute__((__vector_size__(64)));
 typedef long long __m512i __attribute__((__vector_size__(64)));
 
 typedef unsigned char __mmask8;
 typedef unsigned short __mmask16;
 
 /* Rounding mode macros.  */
 #define _MM_FROUND_TO_NEAREST_INT   0x00
 #define _MM_FROUND_TO_NEG_INF       0x01

 
 #define _mm512_maskz_srli_epi64( __U, __A, __B) __extension__ ({ \
 __builtin_ia32_psrlqi512_mask ((__v8di)( __A),( __B),\
               (__v8di)\
               _mm512_setzero_si512 (),\
               (__mmask8)( __U));\
 })
 
+static __inline__ __m512d __DEFAULT_FN_ATTRS
+_mm512_mask_mov_pd (__m512d __W, __mmask8 __U, __m512d __A)
+{
+  return (__m512d) __builtin_ia32_movapd512_mask ((__v8df) __A,
+              (__v8df) __W,
+              (__mmask8) __U);
+}
+
+static __inline__ __m512d __DEFAULT_FN_ATTRS
+_mm512_maskz_mov_pd (__mmask8 __U, __m512d __A)
+{
+  return (__m512d) __builtin_ia32_movapd512_mask ((__v8df) __A,
+              (__v8df)
+              _mm512_setzero_pd (),
+              (__mmask8) __U);
+}
+
+static __inline__ __m512 __DEFAULT_FN_ATTRS
+_mm512_mask_mov_ps (__m512 __W, __mmask16 __U, __m512 __A)
+{
+  return (__m512) __builtin_ia32_movaps512_mask ((__v16sf) __A,
+             (__v16sf) __W,
+             (__mmask16) __U);
+}
+
+static __inline__ __m512 __DEFAULT_FN_ATTRS
+_mm512_maskz_mov_ps (__mmask16 __U, __m512 __A)
+{
+  return (__m512) __builtin_ia32_movaps512_mask ((__v16sf) __A,
+             (__v16sf)
+             _mm512_setzero_ps (),
+             (__mmask16) __U);
+}
+
+
 #undef __DEFAULT_FN_ATTRS
 
 #endif // __AVX512FINTRIN_H

lib/Headers/avx512vlintrin.h

 
 static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm256_cmpeq_epi32_mask(__m256i __a, __m256i __b) {
   return (__mmask8)__builtin_ia32_pcmpeqd256_mask((__v8si)__a, (__v8si)__b,
                                                   (__mmask8)-1);
 }
 
 static __inline__ __mmask8 __DEFAULT_FN_ATTRS_BOTH
 _mm256_mask_cmpeq_epi32_mask(__mmask8 __u, __m256i __a, __m256i __b) {
   return (__mmask8)__builtin_ia32_pcmpeqd256_mask((__v8si)__a, (__v8si)__b,
                                                   __u);
 }
 
 static __inline__ __mmask8 __DEFAULT_FN_ATTRS
 _mm256_cmpeq_epu32_mask(__m256i __a, __m256i __b) {
   return (__mmask8)__builtin_ia32_ucmpd256_mask((__v8si)__a, (__v8si)__b, 0,
                                                 (__mmask8)-1);

 _mm256_maskz_srav_epi64 (__mmask8 __U, __m256i __X, __m256i __Y)
 {
   return (__m256i) __builtin_ia32_psravq256_mask ((__v4di) __X,
               (__v4di) __Y,
               (__v4di)
               _mm256_setzero_si256 (),
               (__mmask8) __U);
 }
+
+static __inline__ __m128d __DEFAULT_FN_ATTRS
+_mm_mask_mov_pd (__m128d __W, __mmask8 __U, __m128d __A)
+{
+  return (__m128d) __builtin_ia32_movapd128_mask ((__v2df) __A,
+              (__v2df) __W,
+              (__mmask8) __U);
+}
+
+static __inline__ __m128d __DEFAULT_FN_ATTRS
+_mm_maskz_mov_pd (__mmask8 __U, __m128d __A)
+{
+  return (__m128d) __builtin_ia32_movapd128_mask ((__v2df) __A,
+              (__v2df)
+              _mm_setzero_pd (),
+              (__mmask8) __U);
+}
+
+static __inline__ __m256d __DEFAULT_FN_ATTRS
+_mm256_mask_mov_pd (__m256d __W, __mmask8 __U, __m256d __A)
+{
+  return (__m256d) __builtin_ia32_movapd256_mask ((__v4df) __A,
+              (__v4df) __W,
+              (__mmask8) __U);
+}
+
+static __inline__ __m256d __DEFAULT_FN_ATTRS
+_mm256_maskz_mov_pd (__mmask8 __U, __m256d __A)
+{
+  return (__m256d) __builtin_ia32_movapd256_mask ((__v4df) __A,
+              (__v4df)
+              _mm256_setzero_pd (),
+              (__mmask8) __U);
+}
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask_mov_ps (__m128 __W, __mmask8 __U, __m128 __A)
+{
+  return (__m128) __builtin_ia32_movaps128_mask ((__v4sf) __A,
+             (__v4sf) __W,
+             (__mmask8) __U);
+}
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_maskz_mov_ps (__mmask8 __U, __m128 __A)
+{
+  return (__m128) __builtin_ia32_movaps128_mask ((__v4sf) __A,
+             (__v4sf)
+             _mm_setzero_ps (),
+             (__mmask8) __U);
+}
+
+static __inline__ __m256 __DEFAULT_FN_ATTRS
+_mm256_mask_mov_ps (__m256 __W, __mmask8 __U, __m256 __A)
+{
+  return (__m256) __builtin_ia32_movaps256_mask ((__v8sf) __A,
+             (__v8sf) __W,
+             (__mmask8) __U);
+}
+
+static __inline__ __m256 __DEFAULT_FN_ATTRS
+_mm256_maskz_mov_ps (__mmask8 __U, __m256 __A)
+{
+  return (__m256) __builtin_ia32_movaps256_mask ((__v8sf) __A,
+             (__v8sf)
+             _mm256_setzero_ps (),
+             (__mmask8) __U);
+}
+
+
 #undef __DEFAULT_FN_ATTRS
 #undef __DEFAULT_FN_ATTRS_BOTH
 
 #endif /* __AVX512VLINTRIN_H */

test/CodeGen/avx512f-builtins.c

 }
 
 __m512i test_mm512_maskz_srli_epi64(__mmask8 __U, __m512i __A) {
   // CHECK-LABEL: @test_mm512_maskz_srli_epi64
   // CHECK: @llvm.x86.avx512.mask.psrl.qi.512
   return _mm512_maskz_srli_epi64(__U, __A, 5); 
 }
 
+__m512d test_mm512_mask_mov_pd(__m512d __W, __mmask8 __U, __m512d __A) {
+  // CHECK-LABEL: @test_mm512_mask_mov_pd
+  // CHECK: @llvm.x86.avx512.mask.mova.pd.512
+  return _mm512_mask_mov_pd(__W, __U, __A); 
+}
+
+__m512d test_mm512_maskz_mov_pd(__mmask8 __U, __m512d __A) {
+  // CHECK-LABEL: @test_mm512_maskz_mov_pd
+  // CHECK: @llvm.x86.avx512.mask.mova.pd.512
+  return _mm512_maskz_mov_pd(__U, __A); 
+}
+
+__m512 test_mm512_mask_mov_ps(__m512 __W, __mmask16 __U, __m512 __A) {
+  // CHECK-LABEL: @test_mm512_mask_mov_ps
+  // CHECK: @llvm.x86.avx512.mask.mova.ps.512
+  return _mm512_mask_mov_ps(__W, __U, __A); 
+}
+
+__m512 test_mm512_maskz_mov_ps(__mmask16 __U, __m512 __A) {
+  // CHECK-LABEL: @test_mm512_maskz_mov_ps
+  // CHECK: @llvm.x86.avx512.mask.mova.ps.512
+  return _mm512_maskz_mov_ps(__U, __A); 
+}

test/CodeGen/avx512vl-builtins.c

 }
 
 __m256i test_mm256_maskz_srav_epi64(__mmask8 __U, __m256i __X, __m256i __Y) {
   // CHECK-LABEL: @test_mm256_maskz_srav_epi64
   // CHECK: @llvm.x86.avx512.mask.psrav.q.256
   return _mm256_maskz_srav_epi64(__U, __X, __Y); 
 }
 
+__m128d test_mm_mask_mov_pd(__m128d __W, __mmask8 __U, __m128d __A) {
+  // CHECK-LABEL: @test_mm_mask_mov_pd
+  // CHECK: @llvm.x86.avx512.mask.mova.pd.128
+  return _mm_mask_mov_pd(__W, __U, __A); 
+}
+
+__m128d test_mm_maskz_mov_pd(__mmask8 __U, __m128d __A) {
+  // CHECK-LABEL: @test_mm_maskz_mov_pd
+  // CHECK: @llvm.x86.avx512.mask.mova.pd.128
+  return _mm_maskz_mov_pd(__U, __A); 
+}
+
+__m256d test_mm256_mask_mov_pd(__m256d __W, __mmask8 __U, __m256d __A) {
+  // CHECK-LABEL: @test_mm256_mask_mov_pd
+  // CHECK: @llvm.x86.avx512.mask.mova.pd.256
+  return _mm256_mask_mov_pd(__W, __U, __A); 
+}
+
+__m256d test_mm256_maskz_mov_pd(__mmask8 __U, __m256d __A) {
+  // CHECK-LABEL: @test_mm256_maskz_mov_pd
+  // CHECK: @llvm.x86.avx512.mask.mova.pd.256
+  return _mm256_maskz_mov_pd(__U, __A); 
+}
+
+__m128 test_mm_mask_mov_ps(__m128 __W, __mmask8 __U, __m128 __A) {
+  // CHECK-LABEL: @test_mm_mask_mov_ps
+  // CHECK: @llvm.x86.avx512.mask.mova.ps.128
+  return _mm_mask_mov_ps(__W, __U, __A); 
+}
+
+__m128 test_mm_maskz_mov_ps(__mmask8 __U, __m128 __A) {
+  // CHECK-LABEL: @test_mm_maskz_mov_ps
+  // CHECK: @llvm.x86.avx512.mask.mova.ps.128
+  return _mm_maskz_mov_ps(__U, __A); 
+}
+
+__m256 test_mm256_mask_mov_ps(__m256 __W, __mmask8 __U, __m256 __A) {
+  // CHECK-LABEL: @test_mm256_mask_mov_ps
+  // CHECK: @llvm.x86.avx512.mask.mova.ps.256
+  return _mm256_mask_mov_ps(__W, __U, __A); 
+}
+
+__m256 test_mm256_maskz_mov_ps(__mmask8 __U, __m256 __A) {
+  // CHECK-LABEL: @test_mm256_maskz_mov_ps
+  // CHECK: @llvm.x86.avx512.mask.mova.ps.256
+  return _mm256_maskz_mov_ps(__U, __A); 
+}
+