maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[X86] Remove masking from the 512-bit floating point max/min builtins. Use select in IR instead. 

llvm-svn: 335200
This commit is contained in:
Craig Topper 2018-06-21 05:01:01 +00:00
parent 296526bf46
commit 2da60bc231
4 changed files with 121 additions and 156 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
clang/include/clang/Basic/BuiltinsX86.def
View File

@ -908,10 +908,10 @@ TARGET_BUILTIN(__builtin_ia32_cvtps2dq512_mask, "V16iV16fV16iUsIi", "nc", "avx51
TARGET_BUILTIN(__builtin_ia32_cvtpd2dq512_mask, "V8iV8dV8iUcIi", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_cvtps2udq512_mask, "V16iV16fV16iUsIi", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_cvtpd2udq512_mask, "V8iV8dV8iUcIi", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_minps512_mask, "V16fV16fV16fV16fUsIi", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_minpd512_mask, "V8dV8dV8dV8dUcIi", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_maxps512_mask, "V16fV16fV16fV16fUsIi", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_maxpd512_mask, "V8dV8dV8dV8dUcIi", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_minps512, "V16fV16fV16fIi", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_minpd512, "V8dV8dV8dIi", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_maxps512, "V16fV16fV16fIi", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_maxpd512, "V8dV8dV8dIi", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_cvtdq2ps512_mask, "V16fV16iV16fUsIi", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_cvtudq2ps512_mask, "V16fV16iV16fUsIi", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_cvtpd2ps512_mask, "V8fV8dV8fUcIi", "nc", "avx512f")

204
clang/lib/Headers/avx512fintrin.h
View File

@ -944,104 +944,78 @@ _mm512_maskz_sub_epi32(__mmask16 __U, __m512i __A, __m512i __B)
(__v16si)_mm512_setzero_si512());
}
#define _mm512_max_round_pd(A, B, R) \
(__m512d)__builtin_ia32_maxpd512((__v8df)(__m512d)(A), \
(__v8df)(__m512d)(B), (int)(R))
#define _mm512_mask_max_round_pd(W, U, A, B, R) \
(__m512d)__builtin_ia32_maxpd512_mask((__v8df)(__m512d)(A), \
(__v8df)(__m512d)(B), \
(__v8df)(__m512d)(W), (__mmask8)(U), \
(int)(R))
(__m512d)__builtin_ia32_selectpd_512((__mmask8)(U), \
(__v8df)_mm512_max_round_pd((A), (B), (R)), \
(__v8df)(W))
#define _mm512_maskz_max_round_pd(U, A, B, R) \
(__m512d)__builtin_ia32_maxpd512_mask((__v8df)(__m512d)(A), \
(__v8df)(__m512d)(B), \
(__v8df)_mm512_setzero_pd(), \
(__mmask8)(U), (int)(R))
#define _mm512_max_round_pd(A, B, R) \
(__m512d)__builtin_ia32_maxpd512_mask((__v8df)(__m512d)(A), \
(__v8df)(__m512d)(B), \
(__v8df)_mm512_undefined_pd(), \
(__mmask8)-1, (int)(R))
(__m512d)__builtin_ia32_selectpd_512((__mmask8)(U), \
(__v8df)_mm512_max_round_pd((A), (B), (R)), \
(__v8df)_mm512_setzero_pd())
static __inline__ __m512d __DEFAULT_FN_ATTRS
_mm512_max_pd(__m512d __A, __m512d __B)
{
return (__m512d) __builtin_ia32_maxpd512_mask ((__v8df) __A,
(__v8df) __B,
(__v8df)
_mm512_setzero_pd (),
(__mmask8) -1,
_MM_FROUND_CUR_DIRECTION);
return (__m512d) __builtin_ia32_maxpd512((__v8df) __A, (__v8df) __B,
_MM_FROUND_CUR_DIRECTION);
}
static __inline__ __m512d __DEFAULT_FN_ATTRS
_mm512_mask_max_pd (__m512d __W, __mmask8 __U, __m512d __A, __m512d __B)
{
return (__m512d) __builtin_ia32_maxpd512_mask ((__v8df) __A,
(__v8df) __B,
(__v8df) __W,
(__mmask8) __U,
_MM_FROUND_CUR_DIRECTION);
return (__m512d)__builtin_ia32_selectpd_512(__U,
(__v8df)_mm512_max_pd(__A, __B),
(__v8df)__W);
}
static __inline__ __m512d __DEFAULT_FN_ATTRS
_mm512_maskz_max_pd (__mmask8 __U, __m512d __A, __m512d __B)
{
return (__m512d) __builtin_ia32_maxpd512_mask ((__v8df) __A,
(__v8df) __B,
(__v8df)
_mm512_setzero_pd (),
(__mmask8) __U,
_MM_FROUND_CUR_DIRECTION);
return (__m512d)__builtin_ia32_selectpd_512(__U,
(__v8df)_mm512_max_pd(__A, __B),
(__v8df)_mm512_setzero_pd());
}
#define _mm512_max_round_ps(A, B, R) \
(__m512)__builtin_ia32_maxps512((__v16sf)(__m512)(A), \
(__v16sf)(__m512)(B), (int)(R))
#define _mm512_mask_max_round_ps(W, U, A, B, R) \
(__m512)__builtin_ia32_maxps512_mask((__v16sf)(__m512)(A), \
(__v16sf)(__m512)(B), \
(__v16sf)(__m512)(W), (__mmask16)(U), \
(int)(R))
(__m512)__builtin_ia32_selectps_512((__mmask16)(U), \
(__v16sf)_mm512_max_round_ps((A), (B), (R)), \
(__v16sf)(W))
#define _mm512_maskz_max_round_ps(U, A, B, R) \
(__m512)__builtin_ia32_maxps512_mask((__v16sf)(__m512)(A), \
(__v16sf)(__m512)(B), \
(__v16sf)_mm512_setzero_ps(), \
(__mmask16)(U), (int)(R))
#define _mm512_max_round_ps(A, B, R) \
(__m512)__builtin_ia32_maxps512_mask((__v16sf)(__m512)(A), \
(__v16sf)(__m512)(B), \
(__v16sf)_mm512_undefined_ps(), \
(__mmask16)-1, (int)(R))
(__m512)__builtin_ia32_selectps_512((__mmask16)(U), \
(__v16sf)_mm512_max_round_ps((A), (B), (R)), \
(__v16sf)_mm512_setzero_ps())
static __inline__ __m512 __DEFAULT_FN_ATTRS
_mm512_max_ps(__m512 __A, __m512 __B)
{
return (__m512) __builtin_ia32_maxps512_mask ((__v16sf) __A,
(__v16sf) __B,
(__v16sf)
_mm512_setzero_ps (),
(__mmask16) -1,
_MM_FROUND_CUR_DIRECTION);
return (__m512) __builtin_ia32_maxps512((__v16sf) __A, (__v16sf) __B,
_MM_FROUND_CUR_DIRECTION);
}
static __inline__ __m512 __DEFAULT_FN_ATTRS
_mm512_mask_max_ps (__m512 __W, __mmask16 __U, __m512 __A, __m512 __B)
{
return (__m512) __builtin_ia32_maxps512_mask ((__v16sf) __A,
(__v16sf) __B,
(__v16sf) __W,
(__mmask16) __U,
_MM_FROUND_CUR_DIRECTION);
return (__m512)__builtin_ia32_selectps_512(__U,
(__v16sf)_mm512_max_ps(__A, __B),
(__v16sf)__W);
}
static __inline__ __m512 __DEFAULT_FN_ATTRS
_mm512_maskz_max_ps (__mmask16 __U, __m512 __A, __m512 __B)
{
return (__m512) __builtin_ia32_maxps512_mask ((__v16sf) __A,
(__v16sf) __B,
(__v16sf)
_mm512_setzero_ps (),
(__mmask16) __U,
_MM_FROUND_CUR_DIRECTION);
return (__m512)__builtin_ia32_selectps_512(__U,
(__v16sf)_mm512_max_ps(__A, __B),
(__v16sf)_mm512_setzero_ps());
}
static __inline__ __m128 __DEFAULT_FN_ATTRS
@ -1205,104 +1179,78 @@ _mm512_maskz_max_epu64 (__mmask8 __M, __m512i __A, __m512i __B)
(__v8di)_mm512_setzero_si512());
}
#define _mm512_min_round_pd(A, B, R) \
(__m512d)__builtin_ia32_minpd512((__v8df)(__m512d)(A), \
(__v8df)(__m512d)(B), (int)(R))
#define _mm512_mask_min_round_pd(W, U, A, B, R) \
(__m512d)__builtin_ia32_minpd512_mask((__v8df)(__m512d)(A), \
(__v8df)(__m512d)(B), \
(__v8df)(__m512d)(W), (__mmask8)(U), \
(int)(R))
(__m512d)__builtin_ia32_selectpd_512((__mmask8)(U), \
(__v8df)_mm512_min_round_pd((A), (B), (R)), \
(__v8df)(W))
#define _mm512_maskz_min_round_pd(U, A, B, R) \
(__m512d)__builtin_ia32_minpd512_mask((__v8df)(__m512d)(A), \
(__v8df)(__m512d)(B), \
(__v8df)_mm512_setzero_pd(), \
(__mmask8)(U), (int)(R))
#define _mm512_min_round_pd(A, B, R) \
(__m512d)__builtin_ia32_minpd512_mask((__v8df)(__m512d)(A), \
(__v8df)(__m512d)(B), \
(__v8df)_mm512_undefined_pd(), \
(__mmask8)-1, (int)(R))
(__m512d)__builtin_ia32_selectpd_512((__mmask8)(U), \
(__v8df)_mm512_min_round_pd((A), (B), (R)), \
(__v8df)_mm512_setzero_pd())
static __inline__ __m512d __DEFAULT_FN_ATTRS
_mm512_min_pd(__m512d __A, __m512d __B)
{
return (__m512d) __builtin_ia32_minpd512_mask ((__v8df) __A,
(__v8df) __B,
(__v8df)
_mm512_setzero_pd (),
(__mmask8) -1,
_MM_FROUND_CUR_DIRECTION);
return (__m512d) __builtin_ia32_minpd512((__v8df) __A, (__v8df) __B,
_MM_FROUND_CUR_DIRECTION);
}
static __inline__ __m512d __DEFAULT_FN_ATTRS
_mm512_mask_min_pd (__m512d __W, __mmask8 __U, __m512d __A, __m512d __B)
{
return (__m512d) __builtin_ia32_minpd512_mask ((__v8df) __A,
(__v8df) __B,
(__v8df) __W,
(__mmask8) __U,
_MM_FROUND_CUR_DIRECTION);
return (__m512d)__builtin_ia32_selectpd_512(__U,
(__v8df)_mm512_min_pd(__A, __B),
(__v8df)__W);
}
#define _mm512_mask_min_round_ps(W, U, A, B, R) \
(__m512)__builtin_ia32_minps512_mask((__v16sf)(__m512)(A), \
(__v16sf)(__m512)(B), \
(__v16sf)(__m512)(W), (__mmask16)(U), \
(int)(R))
#define _mm512_maskz_min_round_ps(U, A, B, R) \
(__m512)__builtin_ia32_minps512_mask((__v16sf)(__m512)(A), \
(__v16sf)(__m512)(B), \
(__v16sf)_mm512_setzero_ps(), \
(__mmask16)(U), (int)(R))
#define _mm512_min_round_ps(A, B, R) \
(__m512)__builtin_ia32_minps512_mask((__v16sf)(__m512)(A), \
(__v16sf)(__m512)(B), \
(__v16sf)_mm512_undefined_ps(), \
(__mmask16)-1, (int)(R))
static __inline__ __m512d __DEFAULT_FN_ATTRS
_mm512_maskz_min_pd (__mmask8 __U, __m512d __A, __m512d __B)
{
return (__m512d) __builtin_ia32_minpd512_mask ((__v8df) __A,
(__v8df) __B,
(__v8df)
_mm512_setzero_pd (),
(__mmask8) __U,
_MM_FROUND_CUR_DIRECTION);
return (__m512d)__builtin_ia32_selectpd_512(__U,
(__v8df)_mm512_min_pd(__A, __B),
(__v8df)_mm512_setzero_pd());
}
#define _mm512_min_round_ps(A, B, R) \
(__m512)__builtin_ia32_minps512((__v16sf)(__m512)(A), \
(__v16sf)(__m512)(B), (int)(R))
#define _mm512_mask_min_round_ps(W, U, A, B, R) \
(__m512)__builtin_ia32_selectps_512((__mmask16)(U), \
(__v16sf)_mm512_min_round_ps((A), (B), (R)), \
(__v16sf)(W))
#define _mm512_maskz_min_round_ps(U, A, B, R) \
(__m512)__builtin_ia32_selectps_512((__mmask16)(U), \
(__v16sf)_mm512_min_round_ps((A), (B), (R)), \
(__v16sf)_mm512_setzero_ps())
static __inline__ __m512 __DEFAULT_FN_ATTRS
_mm512_min_ps(__m512 __A, __m512 __B)
{
return (__m512) __builtin_ia32_minps512_mask ((__v16sf) __A,
(__v16sf) __B,
(__v16sf)
_mm512_setzero_ps (),
(__mmask16) -1,
_MM_FROUND_CUR_DIRECTION);
return (__m512) __builtin_ia32_minps512((__v16sf) __A, (__v16sf) __B,
_MM_FROUND_CUR_DIRECTION);
}
static __inline__ __m512 __DEFAULT_FN_ATTRS
_mm512_mask_min_ps (__m512 __W, __mmask16 __U, __m512 __A, __m512 __B)
{
return (__m512) __builtin_ia32_minps512_mask ((__v16sf) __A,
(__v16sf) __B,
(__v16sf) __W,
(__mmask16) __U,
_MM_FROUND_CUR_DIRECTION);
return (__m512)__builtin_ia32_selectps_512(__U,
(__v16sf)_mm512_min_ps(__A, __B),
(__v16sf)__W);
}
static __inline__ __m512 __DEFAULT_FN_ATTRS
_mm512_maskz_min_ps (__mmask16 __U, __m512 __A, __m512 __B)
{
return (__m512) __builtin_ia32_minps512_mask ((__v16sf) __A,
(__v16sf) __B,
(__v16sf)
_mm512_setzero_ps (),
(__mmask16) __U,
_MM_FROUND_CUR_DIRECTION);
return (__m512)__builtin_ia32_selectps_512(__U,
(__v16sf)_mm512_min_ps(__A, __B),
(__v16sf)_mm512_setzero_ps());
}
static __inline__ __m128 __DEFAULT_FN_ATTRS

10
clang/lib/Sema/SemaChecking.cpp
View File

@ -2298,6 +2298,12 @@ bool Sema::CheckX86BuiltinRoundingOrSAE(unsigned BuiltinID, CallExpr *TheCall) {
case X86::BI__builtin_ia32_vcvttss2usi64:
ArgNum = 1;
break;
case X86::BI__builtin_ia32_maxpd512:
case X86::BI__builtin_ia32_maxps512:
case X86::BI__builtin_ia32_minpd512:
case X86::BI__builtin_ia32_minps512:
ArgNum = 2;
break;
case X86::BI__builtin_ia32_cvtps2pd512_mask:
case X86::BI__builtin_ia32_cvttpd2dq512_mask:
case X86::BI__builtin_ia32_cvttpd2qq512_mask:
@ -2327,12 +2333,8 @@ bool Sema::CheckX86BuiltinRoundingOrSAE(unsigned BuiltinID, CallExpr *TheCall) {
case X86::BI__builtin_ia32_cvtss2sd_round_mask:
case X86::BI__builtin_ia32_getexpsd128_round_mask:
case X86::BI__builtin_ia32_getexpss128_round_mask:
case X86::BI__builtin_ia32_maxpd512_mask:
case X86::BI__builtin_ia32_maxps512_mask:
case X86::BI__builtin_ia32_maxsd_round_mask:
case X86::BI__builtin_ia32_maxss_round_mask:
case X86::BI__builtin_ia32_minpd512_mask:
case X86::BI__builtin_ia32_minps512_mask:
case X86::BI__builtin_ia32_minsd_round_mask:
case X86::BI__builtin_ia32_minss_round_mask:
case X86::BI__builtin_ia32_rcp28sd_round_mask:

55
clang/test/CodeGen/avx512f-builtins.c
View File

@ -8354,140 +8354,155 @@ float test_mm512_cvtss_f32(__m512 A) {
__m512d test_mm512_mask_max_pd (__m512d __W, __mmask8 __U, __m512d __A, __m512d __B)
{
// CHECK-LABEL: @test_mm512_mask_max_pd
// CHECK: @llvm.x86.avx512.mask.max.pd.512
// CHECK: @llvm.x86.avx512.max.pd.512
// CHECK: select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
return _mm512_mask_max_pd (__W,__U,__A,__B);
}
__m512d test_mm512_maskz_max_pd (__mmask8 __U, __m512d __A, __m512d __B)
{
// CHECK-LABEL: @test_mm512_maskz_max_pd
// CHECK: @llvm.x86.avx512.mask.max.pd.512
// CHECK: @llvm.x86.avx512.max.pd.512
// CHECK: select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
return _mm512_maskz_max_pd (__U,__A,__B);
}
__m512 test_mm512_mask_max_ps (__m512 __W, __mmask16 __U, __m512 __A, __m512 __B)
{
// CHECK-LABEL: @test_mm512_mask_max_ps
// CHECK: @llvm.x86.avx512.mask.max.ps.512
// CHECK: @llvm.x86.avx512.max.ps.512
// CHECK: select <16 x i1> %{{.*}}, <16 x float> %{{.*}}, <16 x float> %{{.*}}
return _mm512_mask_max_ps (__W,__U,__A,__B);
}
__m512d test_mm512_mask_max_round_pd(__m512d __W,__mmask8 __U,__m512d __A,__m512d __B)
{
// CHECK-LABEL: @test_mm512_mask_max_round_pd
// CHECK: @llvm.x86.avx512.mask.max.pd.512
// CHECK: @llvm.x86.avx512.max.pd.512
// CHECK: select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
return _mm512_mask_max_round_pd(__W,__U,__A,__B,_MM_FROUND_CUR_DIRECTION);
}
__m512d test_mm512_maskz_max_round_pd(__mmask8 __U,__m512d __A,__m512d __B)
{
// CHECK-LABEL: @test_mm512_maskz_max_round_pd
// CHECK: @llvm.x86.avx512.mask.max.pd.512
// CHECK: @llvm.x86.avx512.max.pd.512
// CHECK: select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
return _mm512_maskz_max_round_pd(__U,__A,__B,_MM_FROUND_CUR_DIRECTION);
}
__m512d test_mm512_max_round_pd(__m512d __A,__m512d __B)
{
// CHECK-LABEL: @test_mm512_max_round_pd
// CHECK: @llvm.x86.avx512.mask.max.pd.512
// CHECK: @llvm.x86.avx512.max.pd.512
return _mm512_max_round_pd(__A,__B,_MM_FROUND_CUR_DIRECTION);
}
__m512 test_mm512_maskz_max_ps (__mmask16 __U, __m512 __A, __m512 __B)
{
// CHECK-LABEL: @test_mm512_maskz_max_ps
// CHECK: @llvm.x86.avx512.mask.max.ps.512
// CHECK: @llvm.x86.avx512.max.ps.512
// CHECK: select <16 x i1> %{{.*}}, <16 x float> %{{.*}}, <16 x float> %{{.*}}
return _mm512_maskz_max_ps (__U,__A,__B);
}
__m512 test_mm512_mask_max_round_ps(__m512 __W,__mmask16 __U,__m512 __A,__m512 __B)
{
// CHECK-LABEL: @test_mm512_mask_max_round_ps
// CHECK: @llvm.x86.avx512.mask.max.ps.512
// CHECK: @llvm.x86.avx512.max.ps.512
// CHECK: select <16 x i1> %{{.*}}, <16 x float> %{{.*}}, <16 x float> %{{.*}}
return _mm512_mask_max_round_ps(__W,__U,__A,__B,_MM_FROUND_CUR_DIRECTION);
}
__m512 test_mm512_maskz_max_round_ps(__mmask16 __U,__m512 __A,__m512 __B)
{
// CHECK-LABEL: @test_mm512_maskz_max_round_ps
// CHECK: @llvm.x86.avx512.mask.max.ps.512
// CHECK: @llvm.x86.avx512.max.ps.512
// CHECK: select <16 x i1> %{{.*}}, <16 x float> %{{.*}}, <16 x float> %{{.*}}
return _mm512_maskz_max_round_ps(__U,__A,__B,_MM_FROUND_CUR_DIRECTION);
}
__m512 test_mm512_max_round_ps(__m512 __A,__m512 __B)
{
// CHECK-LABEL: @test_mm512_max_round_ps
// CHECK: @llvm.x86.avx512.mask.max.ps.512
// CHECK: @llvm.x86.avx512.max.ps.512
return _mm512_max_round_ps(__A,__B,_MM_FROUND_CUR_DIRECTION);
}
__m512d test_mm512_mask_min_pd (__m512d __W, __mmask8 __U, __m512d __A, __m512d __B)
{
// CHECK-LABEL: @test_mm512_mask_min_pd
// CHECK: @llvm.x86.avx512.mask.min.pd.512
// CHECK: @llvm.x86.avx512.min.pd.512
// CHECK: select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
return _mm512_mask_min_pd (__W,__U,__A,__B);
}
__m512d test_mm512_maskz_min_pd (__mmask8 __U, __m512d __A, __m512d __B)
{
// CHECK-LABEL: @test_mm512_maskz_min_pd
// CHECK: @llvm.x86.avx512.mask.min.pd.512
// CHECK: @llvm.x86.avx512.min.pd.512
return _mm512_maskz_min_pd (__U,__A,__B);
}
__m512d test_mm512_mask_min_round_pd(__m512d __W,__mmask8 __U,__m512d __A,__m512d __B)
{
// CHECK-LABEL: @test_mm512_mask_min_round_pd
// CHECK: @llvm.x86.avx512.mask.min.pd.512
// CHECK: @llvm.x86.avx512.min.pd.512
// CHECK: select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
return _mm512_mask_min_round_pd(__W,__U,__A,__B,_MM_FROUND_CUR_DIRECTION);
}
__m512d test_mm512_maskz_min_round_pd(__mmask8 __U,__m512d __A,__m512d __B)
{
// CHECK-LABEL: @test_mm512_maskz_min_round_pd
// CHECK: @llvm.x86.avx512.mask.min.pd.512
// CHECK: @llvm.x86.avx512.min.pd.512
// CHECK: select <8 x i1> %{{.*}}, <8 x double> %{{.*}}, <8 x double> %{{.*}}
return _mm512_maskz_min_round_pd(__U,__A,__B,_MM_FROUND_CUR_DIRECTION);
}
__m512d test_mm512_min_round_pd( __m512d __A,__m512d __B)
{
// CHECK-LABEL: @test_mm512_min_round_pd
// CHECK: @llvm.x86.avx512.mask.min.pd.512
// CHECK: @llvm.x86.avx512.min.pd.512
return _mm512_min_round_pd(__A,__B,_MM_FROUND_CUR_DIRECTION);
}
__m512 test_mm512_mask_min_ps (__m512 __W, __mmask16 __U, __m512 __A, __m512 __B)
{
// CHECK-LABEL: @test_mm512_mask_min_ps
// CHECK: @llvm.x86.avx512.mask.min.ps.512
// CHECK: @llvm.x86.avx512.min.ps.512
// CHECK: select <16 x i1> %{{.*}}, <16 x float> %{{.*}}, <16 x float> %{{.*}}
return _mm512_mask_min_ps (__W,__U,__A,__B);
}
__m512 test_mm512_maskz_min_ps (__mmask16 __U, __m512 __A, __m512 __B)
{
// CHECK-LABEL: @test_mm512_maskz_min_ps
// CHECK: @llvm.x86.avx512.mask.min.ps.512
// CHECK: @llvm.x86.avx512.min.ps.512
// CHECK: select <16 x i1> %{{.*}}, <16 x float> %{{.*}}, <16 x float> %{{.*}}
return _mm512_maskz_min_ps (__U,__A,__B);
}
__m512 test_mm512_mask_min_round_ps(__m512 __W,__mmask16 __U,__m512 __A,__m512 __B)
{
// CHECK-LABEL: @test_mm512_mask_min_round_ps
// CHECK: @llvm.x86.avx512.mask.min.ps.512
// CHECK: @llvm.x86.avx512.min.ps.512
// CHECK: select <16 x i1> %{{.*}}, <16 x float> %{{.*}}, <16 x float> %{{.*}}
return _mm512_mask_min_round_ps(__W,__U,__A,__B,_MM_FROUND_CUR_DIRECTION);
}
__m512 test_mm512_maskz_min_round_ps(__mmask16 __U,__m512 __A,__m512 __B)
{
// CHECK-LABEL: @test_mm512_maskz_min_round_ps
// CHECK: @llvm.x86.avx512.mask.min.ps.512
// CHECK: @llvm.x86.avx512.min.ps.512
// CHECK: select <16 x i1> %{{.*}}, <16 x float> %{{.*}}, <16 x float> %{{.*}}
return _mm512_maskz_min_round_ps(__U,__A,__B,_MM_FROUND_CUR_DIRECTION);
}
__m512 test_mm512_min_round_ps(__m512 __A,__m512 __B)
{
// CHECK-LABEL: @test_mm512_min_round_ps
// CHECK: @llvm.x86.avx512.mask.min.ps.512
// CHECK: @llvm.x86.avx512.min.ps.512
return _mm512_min_round_ps(__A,__B,_MM_FROUND_CUR_DIRECTION);
}