Skip to content

Commit

Permalink
[Clang][Builtin][AVX512] Adding intrinsics for VGETMANT{PD|PS} and VG…
Browse files Browse the repository at this point in the history
…ETEXP{PD|PS} instruction set

Differential Revision: http://reviews.llvm.org/D19197

llvm-svn: 266763
  • Loading branch information
Michael Zuckerman authored and Michael Zuckerman committed Apr 19, 2016
1 parent a4810fa commit 6fa512c
Show file tree
Hide file tree
Showing 5 changed files with 481 additions and 0 deletions.
8 changes: 8 additions & 0 deletions clang/include/clang/Basic/BuiltinsX86.def
Expand Up @@ -2121,6 +2121,14 @@ TARGET_BUILTIN(__builtin_ia32_insertf64x2_256_mask, "V4dV4dV2dIiV4dUc","","avx51
TARGET_BUILTIN(__builtin_ia32_inserti64x2_256_mask, "V4LLiV4LLiV2LLiIiV4LLiUc","","avx512dq,avx512vl")
TARGET_BUILTIN(__builtin_ia32_insertf32x4_256_mask, "V8fV8fV4fIiV8fUc","","avx512vl")
TARGET_BUILTIN(__builtin_ia32_inserti32x4_256_mask, "V8iV8iV4iIiV8iUc","","avx512vl")
TARGET_BUILTIN(__builtin_ia32_getmantpd128_mask, "V2dV2diV2dUc","","avx512vl")
TARGET_BUILTIN(__builtin_ia32_getmantpd256_mask, "V4dV4diV4dUc","","avx512vl")
TARGET_BUILTIN(__builtin_ia32_getmantps128_mask, "V4fV4fiV4fUc","","avx512vl")
TARGET_BUILTIN(__builtin_ia32_getmantps256_mask, "V8fV8fiV8fUc","","avx512vl")
TARGET_BUILTIN(__builtin_ia32_getmantpd512_mask, "V8dV8diV8dUcIi","","avx512f")
TARGET_BUILTIN(__builtin_ia32_getmantps512_mask, "V16fV16fiV16fUsIi","","avx512f")
TARGET_BUILTIN(__builtin_ia32_getexppd512_mask, "V8dV8dV8dUcIi","","avx512f")
TARGET_BUILTIN(__builtin_ia32_getexpps512_mask, "V16fV16fV16fUsIi","","avx512f")

#undef BUILTIN
#undef TARGET_BUILTIN
173 changes: 173 additions & 0 deletions clang/lib/Headers/avx512fintrin.h
Expand Up @@ -6148,6 +6148,179 @@ __builtin_ia32_inserti64x4_mask ((__v8di)( __A),\
(__mmask8)( __U));\
})

#define _mm512_getmant_round_pd( __A, __B, __C, __R) __extension__ ({ \
__builtin_ia32_getmantpd512_mask ((__v8df)( __A),\
(__C << 2) |( __B),\
(__v8df) _mm512_undefined_pd (),\
(__mmask8) -1,( __R));\
})

#define _mm512_mask_getmant_round_pd( __W, __U, __A, __B, __C, __R) __extension__ ({ \
__builtin_ia32_getmantpd512_mask ((__v8df)( __A),\
(__C << 2) |( __B),\
(__v8df)( __W),(__mmask8)( __U),\
( __R));\
})

#define _mm512_maskz_getmant_round_pd( __U, __A, __B, __C, __R) __extension__ ({ \
__builtin_ia32_getmantpd512_mask ((__v8df)( __A),\
(__C << 2) |( __B),\
(__v8df) _mm512_setzero_pd (),\
(__mmask8)( __U),( __R));\
})

#define _mm512_getmant_pd( __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantpd512_mask ((__v8df)( __A),\
(__C << 2) |( __B),\
(__v8df) _mm512_setzero_pd (),\
(__mmask8) -1, _MM_FROUND_CUR_DIRECTION);\
})

#define _mm512_mask_getmant_pd( __W, __U, __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantpd512_mask ((__v8df)( __A),\
(__C << 2) |( __B),\
(__v8df)( __W), (__mmask8)( __U), _MM_FROUND_CUR_DIRECTION);\
})

#define _mm512_maskz_getmant_pd( __U, __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantpd512_mask ((__v8df)( __A),\
(__C << 2) |( __B),\
(__v8df) _mm512_setzero_pd (),\
(__mmask8)( __U), _MM_FROUND_CUR_DIRECTION);\
})

#define _mm512_getmant_round_ps( __A, __B, __C, __R) __extension__ ({ \
__builtin_ia32_getmantps512_mask ((__v16sf)( __A),\
(__C << 2) |( __B),\
(__v16sf) _mm512_undefined_ps (),\
(__mmask16) -1,( __R));\
})

#define _mm512_mask_getmant_round_ps( __W, __U, __A, __B, __C, __R) __extension__ ({ \
__builtin_ia32_getmantps512_mask ((__v16sf)( __A),\
(__C << 2) |( __B),\
(__v16sf)( __W),(__mmask16)( __U),\
( __R));\
})

#define _mm512_maskz_getmant_round_ps( __U, __A, __B, __C, __R) __extension__ ({ \
__builtin_ia32_getmantps512_mask ((__v16sf)( __A),\
(__C << 2) |( __B),\
(__v16sf) _mm512_setzero_ps (),\
( __U),( __R));\
})

#define _mm512_getmant_ps( __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantps512_mask ((__v16sf)( __A),\
(__C << 2) |( __B),\
(__v16sf) _mm512_undefined_ps (),\
(__mmask16) -1, _MM_FROUND_CUR_DIRECTION);\
})

#define _mm512_mask_getmant_ps( __W, __U, __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantps512_mask ((__v16sf)( __A),\
(__C << 2) |( __B),\
(__v16sf)( __W),(__mmask16) ( __U),\
_MM_FROUND_CUR_DIRECTION);\
})

#define _mm512_maskz_getmant_ps( __U, __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantps512_mask ((__v16sf)( __A),\
(__C << 2) |( __B),\
(__v16sf) _mm512_setzero_ps (),\
(__mmask16)( __U),_MM_FROUND_CUR_DIRECTION);\
})

#define _mm512_getexp_round_pd( __A, __R) __extension__ ({ \
__builtin_ia32_getexppd512_mask ((__v8df)( __A),\
(__v8df) _mm512_undefined_pd (),\
(__mmask8) -1,( __R));\
})

#define _mm512_mask_getexp_round_pd( __W, __U, __A, __R) __extension__ ({ \
__builtin_ia32_getexppd512_mask ((__v8df)( __A),\
(__v8df)( __W),\
(__mmask8)( __U),( __R));\
})

#define _mm512_maskz_getexp_round_pd( __U, __A, __R) __extension__ ({ \
__builtin_ia32_getexppd512_mask ((__v8df)( __A),\
(__v8df) _mm512_setzero_pd (),\
(__mmask8)( __U),( __R));\
})

static __inline__ __m512d __DEFAULT_FN_ATTRS
_mm512_getexp_pd (__m512d __A)
{
return (__m512d) __builtin_ia32_getexppd512_mask ((__v8df) __A,
(__v8df) _mm512_undefined_pd (),
(__mmask8) -1,
_MM_FROUND_CUR_DIRECTION);
}

static __inline__ __m512d __DEFAULT_FN_ATTRS
_mm512_mask_getexp_pd (__m512d __W, __mmask8 __U, __m512d __A)
{
return (__m512d) __builtin_ia32_getexppd512_mask ((__v8df) __A,
(__v8df) __W,
(__mmask8) __U,
_MM_FROUND_CUR_DIRECTION);
}

static __inline__ __m512d __DEFAULT_FN_ATTRS
_mm512_maskz_getexp_pd (__mmask8 __U, __m512d __A)
{
return (__m512d) __builtin_ia32_getexppd512_mask ((__v8df) __A,
(__v8df) _mm512_setzero_pd (),
(__mmask8) __U,
_MM_FROUND_CUR_DIRECTION);
}

#define _mm512_getexp_round_ps( __A, __R) __extension__ ({ \
__builtin_ia32_getexpps512_mask ((__v16sf)( __A),\
(__v16sf) _mm512_undefined_ps (),\
(__mmask16) -1,( __R));\
})

#define _mm512_mask_getexp_round_ps( __W, __U, __A, __R) __extension__ ({ \
__builtin_ia32_getexpps512_mask ((__v16sf)( __A),\
(__v16sf)( __W),\
(__mmask16)( __U),( __R));\
})

#define _mm512_maskz_getexp_round_ps( __U, __A, __R) __extension__ ({ \
__builtin_ia32_getexpps512_mask ((__v16sf)( __A),\
(__v16sf) _mm512_setzero_ps (),\
(__mmask16)( __U),( __R));\
})

static __inline__ __m512 __DEFAULT_FN_ATTRS
_mm512_getexp_ps (__m512 __A)
{
return (__m512) __builtin_ia32_getexpps512_mask ((__v16sf) __A,
(__v16sf) _mm512_undefined_ps (),
(__mmask16) -1,
_MM_FROUND_CUR_DIRECTION);
}

static __inline__ __m512 __DEFAULT_FN_ATTRS
_mm512_mask_getexp_ps (__m512 __W, __mmask16 __U, __m512 __A)
{
return (__m512) __builtin_ia32_getexpps512_mask ((__v16sf) __A,
(__v16sf) __W,
(__mmask16) __U,
_MM_FROUND_CUR_DIRECTION);
}

static __inline__ __m512 __DEFAULT_FN_ATTRS
_mm512_maskz_getexp_ps (__mmask16 __U, __m512 __A)
{
return (__m512) __builtin_ia32_getexpps512_mask ((__v16sf) __A,
(__v16sf) _mm512_setzero_ps (),
(__mmask16) __U,
_MM_FROUND_CUR_DIRECTION);
}

#undef __DEFAULT_FN_ATTRS

#endif // __AVX512FINTRIN_H
84 changes: 84 additions & 0 deletions clang/lib/Headers/avx512vlintrin.h
Expand Up @@ -8894,6 +8894,90 @@ __builtin_ia32_inserti32x4_256_mask ((__v8si)( __A),\
(__mmask8) ( __U));\
})

#define _mm_getmant_pd( __A, __B, __C) __extension__({\
__builtin_ia32_getmantpd128_mask ((__v2df) __A,\
(__C << 2) | __B,\
(__v2df) _mm_setzero_pd (),\
(__mmask8) -1);\
})

#define _mm_mask_getmant_pd( __W, __U, __A, __B, __C) __extension__({\
__builtin_ia32_getmantpd128_mask ((__v2df) __A,\
(__C << 2) | __B,\
(__v2df) __W,\
(__mmask8) __U);\
})

#define _mm_maskz_getmant_pd( __U, __A, __B, __C) __extension__({\
__builtin_ia32_getmantpd128_mask ((__v2df) __A,\
(__C << 2) | __B,\
(__v2df) _mm_setzero_pd (),\
(__mmask8) __U);\
})

#define _mm256_getmant_pd( __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantpd256_mask ((__v4df)( __A),\
(__C << 2) |( __B),\
(__v4df) _mm256_setzero_pd (),\
(__mmask8) -1);\
})

#define _mm256_mask_getmant_pd( __W, __U, __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantpd256_mask ((__v4df)( __A),\
(__C << 2) |( __B),\
(__v4df)( __W),\
(__mmask8)( __U));\
})

#define _mm256_maskz_getmant_pd( __U, __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantpd256_mask ((__v4df)( __A),\
(__C << 2) |( __B),\
(__v4df) _mm256_setzero_pd (),\
(__mmask8)( __U));\
})

#define _mm_getmant_ps( __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantps128_mask ((__v4sf)( __A),\
(__C << 2) |( __B),\
(__v4sf) _mm_setzero_ps (),\
(__mmask8) -1);\
})

#define _mm_mask_getmant_ps( __W, __U, __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantps128_mask ((__v4sf)( __A),\
(__C << 2) |( __B),\
(__v4sf)( __W),\
(__mmask8)( __U));\
})

#define _mm_maskz_getmant_ps( __U, __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantps128_mask ((__v4sf)( __A),\
(__C << 2) |( __B),\
(__v4sf) _mm_setzero_ps (),\
(__mmask8)( __U));\
})

#define _mm256_getmant_ps( __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantps256_mask ((__v8sf)( __A),\
(__C << 2) |( __B),\
(__v8sf) _mm256_setzero_ps (),\
(__mmask8) -1);\
})

#define _mm256_mask_getmant_ps( __W, __U, __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantps256_mask ((__v8sf)( __A),\
(__C << 2) |( __B),\
(__v8sf)( __W),\
(__mmask8)( __U));\
})

#define _mm256_maskz_getmant_ps( __U, __A, __B, __C) __extension__ ({ \
__builtin_ia32_getmantps256_mask ((__v8sf)( __A),\
(__C << 2) |( __B),\
(__v8sf) _mm256_setzero_ps (),\
(__mmask8)( __U));\
})

#undef __DEFAULT_FN_ATTRS
#undef __DEFAULT_FN_ATTRS_BOTH

Expand Down

0 comments on commit 6fa512c

Please sign in to comment.