[CLANG][AVX512][BUILTIN] Completing and Adding scalar Intrinsics to clang

[m_zuckerman]

Summary

rsqrt14s{s|d}
rcp14s{s|d}
getexps{s|d}
getmants{s|d}
sqrts{s|d}
fmadd_s{s|d}
fmadd_round_s{s|d}
fmsub_s{s|d}
fmsub_round_s{s|d}
fnmadd_s{s|d}
fnmadd_round_s{s|d}
fnmsub_s{s|d}
fnmsub_round_s{s|d}

[m_zuckerman]

Ping

[delena]

you mixed all together. Please add fmaddss, fmadsd only.

[AsafBadouh]

lib/Headers/avx512fintrin.h

4356

I think it should be (v4sf) (-(B))

4364

same
I think it should be (v4sf) (-(B))

4409

please add round parentheses to __A

4479

some "-" are missing

4571

some "-" is missing

4579

some "-" is missing

4660

I'm not sure but I think it should be:
(v2df) (A),
(v2df) -(B),
(v2df) -(W),

[m_zuckerman]

This review was split into two reviews:

The first, [Clang][AVX512][BUILTIN] Adding scalar intrinsics for rsqrt14 ,rcp14, getexp and getmant instruction set
http://reviews.llvm.org/D19326
The second, [Clang][AVX512][Builtin] adding intrinsics for vf{n}madd{ss|sd} and vf{n}sub{ss|sd} instruction set
http://reviews.llvm.org/D19320

include/clang/Basic/BuiltinsX86.def

 TARGET_BUILTIN(__builtin_ia32_loaddqudi128_mask, "V2LLiV2LLi*V2LLiUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_loaddqudi256_mask, "V4LLiV4LLi*V4LLiUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_loaddqusi128_mask, "V4iV4i*V4iUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_loaddqusi256_mask, "V8iV8i*V8iUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_loadupd128_mask, "V2dV2d*V2dUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_loadupd256_mask, "V4dV4d*V4dUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_loadups128_mask, "V4fV4f*V4fUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_loadups256_mask, "V8fV8f*V8fUc","","avx512vl")
+TARGET_BUILTIN(__builtin_ia32_sqrtsd_round_mask, "V2dV2dV2dV2dUcIi", "", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_sqrtss_round_mask, "V4fV4fV4fV4fUcIi", "", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_vfmaddss3_mask,  "V4fV4fV4fV4fUcIi", "", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_vfmaddss3_maskz, "V4fV4fV4fV4fUcIi", "", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_vfmaddss3_mask3, "V4fV4fV4fV4fUcIi", "", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_vfmaddsd3_mask,  "V2dV2dV2dV2dUcIi", "", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_vfmaddsd3_maskz, "V2dV2dV2dV2dUcIi", "", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_vfmaddsd3_mask3, "V2dV2dV2dV2dUcIi", "", "avx512f")
 
 #undef BUILTIN
 #undef TARGET_BUILTIN

lib/Headers/avx512fintrin.h

 {
   return (__m128) __builtin_ia32_rsqrt14ss_mask ((__v4sf) __A,
              (__v4sf) __B,
              (__v4sf)
              _mm_setzero_ps (),
              (__mmask8) -1);
 }
 
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask_rsqrt14_ss (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B) 
+{
+ return (__m128) __builtin_ia32_rsqrt14ss_mask ((__v4sf) __A,
+          (__v4sf) __B,
+          (__v4sf) __W,
+          (__mmask8) __U);
+}
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_maskz_rsqrt14_ss (__mmask8 __U, __m128 __A, __m128 __B) 
+{
+ return (__m128) __builtin_ia32_rsqrt14ss_mask ((__v4sf) __A,
+          (__v4sf) __B,
+          (__v4sf) _mm_setzero_ps (),
+          (__mmask8) __U);
+}
+
 static  __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_rsqrt14_sd(__m128d __A, __m128d __B)
 {
   return (__m128d) __builtin_ia32_rsqrt14sd_mask ((__v2df) __A,
               (__v2df) __B,
               (__v2df)
               _mm_setzero_pd (),
               (__mmask8) -1);
 }
 
+static __inline__ __m128d __DEFAULT_FN_ATTRS
+_mm_mask_rsqrt14_sd (__m128d __W, __mmask8 __U, __m128d __A, __m128d __B) 
+{
+ return (__m128d) __builtin_ia32_rsqrt14sd_mask ( (__v2df) __A,
+          (__v2df) __B,
+          (__v2df) __W,
+          (__mmask8) __U);
+}
+
+static __inline__ __m128d __DEFAULT_FN_ATTRS
+_mm_maskz_rsqrt14_sd (__mmask8 __U, __m128d __A, __m128d __B) 
+{
+ return (__m128d) __builtin_ia32_rsqrt14sd_mask ( (__v2df) __A,
+          (__v2df) __B,
+          (__v2df) _mm_setzero_pd (),
+          (__mmask8) __U);
+}
+
 static  __inline__ __m512d __DEFAULT_FN_ATTRS
 _mm512_rcp14_pd(__m512d __A)
 {
   return (__m512d) __builtin_ia32_rcp14pd512_mask ((__v8df) __A,
                (__v8df)
                _mm512_setzero_pd (),
                (__mmask8) -1);
 }

 {
   return (__m128) __builtin_ia32_rcp14ss_mask ((__v4sf) __A,
                  (__v4sf) __B,
                  (__v4sf)
                  _mm_setzero_ps (),
                  (__mmask8) -1);
 }
 
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask_rcp14_ss (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B) 
+{
+ return (__m128) __builtin_ia32_rcp14ss_mask ((__v4sf) __A,
+          (__v4sf) __B,
+          (__v4sf) __W,
+          (__mmask8) __U);
+}
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_maskz_rcp14_ss (__mmask8 __U, __m128 __A, __m128 __B) 
+{
+ return (__m128) __builtin_ia32_rcp14ss_mask ((__v4sf) __A,
+          (__v4sf) __B,
+          (__v4sf) _mm_setzero_ps (),
+          (__mmask8) __U);
+}
+
 static  __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_rcp14_sd(__m128d __A, __m128d __B)
 {
   return (__m128d) __builtin_ia32_rcp14sd_mask ((__v2df) __A,
             (__v2df) __B,
             (__v2df)
             _mm_setzero_pd (),
             (__mmask8) -1);
 }
 
+static __inline__ __m128d __DEFAULT_FN_ATTRS
+_mm_mask_rcp14_sd (__m128d __W, __mmask8 __U, __m128d __A, __m128d __B) 
+{
+ return (__m128d) __builtin_ia32_rcp14sd_mask ( (__v2df) __A,
+          (__v2df) __B,
+          (__v2df) __W,
+          (__mmask8) __U);
+}
+
+static __inline__ __m128d __DEFAULT_FN_ATTRS
+_mm_maskz_rcp14_sd (__mmask8 __U, __m128d __A, __m128d __B) 
+{
+ return (__m128d) __builtin_ia32_rcp14sd_mask ( (__v2df) __A,
+          (__v2df) __B,
+          (__v2df) _mm_setzero_pd (),
+          (__mmask8) __U);
+}
+
 static __inline __m512 __DEFAULT_FN_ATTRS
 _mm512_floor_ps(__m512 __A)
 {
   return (__m512) __builtin_ia32_rndscaleps_mask ((__v16sf) __A,
                                                   _MM_FROUND_FLOOR,
                                                   (__v16sf) __A, -1,
                                                   _MM_FROUND_CUR_DIRECTION);
 }

 
 static __inline__ __m128d __DEFAULT_FN_ATTRS
 _mm_getexp_sd (__m128d __A, __m128d __B)
 {
   return (__m128d) __builtin_ia32_getexpsd128_round_mask ((__v2df) __A,
                  (__v2df) __B, (__v2df) _mm_setzero_pd(), (__mmask8) -1, _MM_FROUND_CUR_DIRECTION);
 }
 
+static __inline__ __m128d __DEFAULT_FN_ATTRS
+_mm_mask_getexp_sd (__m128d __W, __mmask8 __U, __m128d __A, __m128d __B) 
+{
+ return (__m128d) __builtin_ia32_getexpsd128_round_mask ( (__v2df) __A,
+          (__v2df) __B,
+          (__v2df) __W,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask_getexp_round_sd( __W, __U, __A, __B, __R) __extension__ ({\
+__builtin_ia32_getexpsd128_round_mask ((__v2df) __A,\
+          (__v2df) __B,\
+          (__v2df) __W,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128d __DEFAULT_FN_ATTRS
+_mm_maskz_getexp_sd (__mmask8 __U, __m128d __A, __m128d __B) 
+{
+ return (__m128d) __builtin_ia32_getexpsd128_round_mask ( (__v2df) __A,
+          (__v2df) __B,
+          (__v2df) _mm_setzero_pd (),
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_maskz_getexp_round_sd( __U, __A, __B, __R) __extension__ ({\
+__builtin_ia32_getexpsd128_round_mask ( (__v2df) __A,\
+          (__v2df) __B,\
+          (__v2df) _mm_setzero_pd (),\
+          (__mmask8) __U,\
+          __R);\
+})
+
 #define _mm_getexp_round_ss( __A, __B, __R) __extension__ ({ \
 __builtin_ia32_getexpss128_round_mask ((__v4sf)( __A),\
                 (__v4sf)( __B), (__v4sf)  _mm_setzero_ps(), (__mmask8) -1,\
                 ( __R));\
 })
 
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_getexp_ss (__m128 __A, __m128 __B)
 {
   return (__m128) __builtin_ia32_getexpss128_round_mask ((__v4sf) __A,
                 (__v4sf) __B, (__v4sf)  _mm_setzero_ps(), (__mmask8) -1, _MM_FROUND_CUR_DIRECTION);
 }
 
+static __inline__ __m128d __DEFAULT_FN_ATTRS
+_mm_mask_getexp_ss (__m128d __W, __mmask8 __U, __m128d __A, __m128d __B) 
+{
+ return (__m128d) __builtin_ia32_getexpss128_round_mask ((__v4sf) __A,
+          (__v4sf) __B,
+          (__v4sf) __W,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask_getexp_round_ss( __W, __U, __A, __B, __R) __extension__ ({\
+__builtin_ia32_getexpss128_round_mask ((__v4sf) __A,\
+          (__v4sf) __B,\
+          (__v4sf) __W,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128d __DEFAULT_FN_ATTRS
+_mm_maskz_getexp_ss (__mmask8 __U, __m128d __A, __m128d __B) 
+{
+ return (__m128d) __builtin_ia32_getexpss128_round_mask ((__v4sf) __A,
+          (__v4sf) __B,
+          (__v4sf) _mm_setzero_pd (),
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_maskz_getexp_round_ss( __U, __A, __B, __R) __extension__ ({\
+__builtin_ia32_getexpss128_round_mask ((__v4sf) __A,\
+          (__v4sf) __B,\
+          (__v4sf) _mm_setzero_ps (),\
+          (__mmask8) __U,\
+          __R);\
+})
+
 #define _mm_getmant_round_sd( __A, __B, __C, __D, __R) __extension__ ({ \
 __builtin_ia32_getmantsd_round_mask ((__v2df)( __A),\
               (__v2df)( __B),\
               (( __D) << 2) |( __C), (__v2df) _mm_setzero_pd(), (__mmask8) -1,\
               ( __R));\
 })
 
 #define _mm_getmant_sd( __A,  __B,  __C, __D)  __extension__ ({ \
 __builtin_ia32_getmantsd_round_mask ((__v2df)( __A),\
               (__v2df)( __B),\
               (( __D) << 2) |( __C), (__v2df) _mm_setzero_pd(), (__mmask8) -1,\
               _MM_FROUND_CUR_DIRECTION);\
 })
 
+#define _mm_mask_getmant_sd( __W, __U, __A, __B, __C, __D) __extension__ ({\
+__builtin_ia32_getmantsd_round_mask ( (__v2df) __A,\
+          (__v2df) __B,\
+          (( __D) << 2) |( __C),\
+          (__v2df) __W,\
+          (__mmask8) __U,\
+          _MM_FROUND_CUR_DIRECTION);\
+})
+
+#define _mm_mask_getmant_round_sd( __W, __U, __A, __B, __C, __D, __R)({\
+__builtin_ia32_getmantsd_round_mask ( (__v2df) __A,\
+          (__v2df) __B,\
+          (( __D) << 2) |( __C),\
+          (__v2df) __W,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+#define _mm_maskz_getmant_sd( __U, __A, __B, __C, __D) __extension__ ({\
+__builtin_ia32_getmantsd_round_mask ( (__v2df) __A,\
+          (__v2df) __B,\
+          (( __D) << 2) |( __C),\
+          (__v2df) _mm_setzero_pd (),\
+          (__mmask8) __U,\
+          _MM_FROUND_CUR_DIRECTION);\
+})
+
+#define _mm_maskz_getmant_round_sd( __U, __A, __B, __C, __D, __R) __extension__ ({\
+__builtin_ia32_getmantsd_round_mask ( (__v2df) __A,\
+          (__v2df) __B,\
+          (( __D) << 2) |( __C),\
+          (__v2df) _mm_setzero_pd (),\
+          (__mmask8) __U,\
+          __R);\
+})
+
 #define _mm_getmant_round_ss( __A, __B, __C, __D, __R) __extension__ ({ \
 __builtin_ia32_getmantss_round_mask ((__v4sf)( __A),\
               (__v4sf)( __B),\
               ((__D) << 2) |( __C), (__v4sf)  _mm_setzero_ps(), (__mmask8) -1,\
               ( __R));\
 })
 
 #define  _mm_getmant_ss(__A,  __B,  __C, __D) __extension__ ({ \
 __builtin_ia32_getmantss_round_mask ((__v4sf)( __A),\
               (__v4sf)( __B),\
               ((__D) << 2) |( __C), (__v4sf)  _mm_setzero_ps(), (__mmask8) -1,\
               _MM_FROUND_CUR_DIRECTION);\
 })
 
+#define _mm_mask_getmant_ss( __W, __U, __A, __B, __C, __D) __extension__ ({\
+__builtin_ia32_getmantss_round_mask ((__v4sf) __A,\
+          (__v4sf) __B,\
+          (( __D) << 2) |( __C),\
+          (__v4sf) __W,\
+          (__mmask8) __U,\
+          _MM_FROUND_CUR_DIRECTION);\
+})
+
+#define _mm_mask_getmant_round_ss( __W, __U, __A, __B, __C, __D, __R)({\
+__builtin_ia32_getmantss_round_mask ((__v4sf) __A,\
+          (__v4sf) __B,\
+          (( __D) << 2) |( __C),\
+          (__v4sf) __W,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+#define _mm_maskz_getmant_ss( __U, __A, __B, __C, __D) __extension__ ({\
+__builtin_ia32_getmantss_round_mask ((__v4sf) __A,\
+          (__v4sf) __B,\
+          (( __D) << 2) |( __C),\
+          (__v4sf) _mm_setzero_pd (),\
+          (__mmask8) __U,\
+          _MM_FROUND_CUR_DIRECTION);\
+})
+
+#define _mm_maskz_getmant_round_ss( __U, __A, __B, __C, __D, __R) __extension__ ({\
+__builtin_ia32_getmantss_round_mask ((__v4sf) __A,\
+          (__v4sf) __B,\
+          (( __D) << 2) |( __C),\
+          (__v4sf) _mm_setzero_ps (),\
+          (__mmask8) __U,\
+          __R);\
+})
 
 static __inline__ __mmask16 __DEFAULT_FN_ATTRS
 _mm512_kmov (__mmask16 __A)
 {
   return  __A;
 }
 
 #define _mm_comi_round_sd(__A, __B, __P, __R) __extension__ ({\
 __builtin_ia32_vcomisd ((__v2df) (__A), (__v2df) (__B), ( __P), (  __R));\
 })
 
 #define _mm_comi_round_ss( __A,  __B, __P, __R) __extension__ ({\
 __builtin_ia32_vcomiss ((__v4sf) (__A), (__v4sf) (__B), ( __P), ( __R));\
 })
 
+#define _mm_sqrt_round_ss( __A, __B, __R) __extension__ ({\
+__builtin_ia32_sqrtss_round_mask ((__v4sf) __A,\
+        (__v4sf) __B,\
+        (__v4sf) _mm_setzero_ps (),\
+        (__mmask8) -1,\
+        __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask_sqrt_ss (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B) 
+{
+ return (__m128) __builtin_ia32_sqrtss_round_mask ((__v4sf) __A,
+          (__v4sf) __B,
+          (__v4sf) __W,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask_sqrt_round_ss( __W, __U, __A, __B, __R) __extension__ ({\
+__builtin_ia32_sqrtss_round_mask ((__v4sf) __A,\
+        (__v4sf) __B,\
+        (__v4sf) __W,\
+        (__mmask8) __U,\
+        __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_maskz_sqrt_ss (__mmask8 __U, __m128 __A, __m128 __B) 
+{
+return (__m128) __builtin_ia32_sqrtss_round_mask ((__v4sf) __A,
+          (__v4sf) __B,
+          (__v4sf) _mm_setzero_ps(),
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_maskz_sqrt_round_ss( __U, __A, __B, __R) __extension__ ({\
+__builtin_ia32_sqrtss_round_mask ((__v4sf) __A,\
+        (__v4sf) __B,\
+        (__v4sf) _mm_setzero_ps(),\
+        (__mmask8) __U,\
+        __R);\
+})
+
+#define _mm_sqrt_round_sd( __A, __B, __R) __extension__ ({\
+__builtin_ia32_sqrtsd_round_mask ((__v2df) __A,\
+        (__v2df) __B,\
+        (__v2df) _mm_setzero_pd (),\
+        (__mmask8) -1,\
+        __R);\
+})
+
+static __inline__ __m128d __DEFAULT_FN_ATTRS
+_mm_mask_sqrt_sd (__m128d __W, __mmask8 __U, __m128d __A, __m128d __B) 
+{
+ return (__m128d) __builtin_ia32_sqrtsd_round_mask ( (__v2df) __A,
+          (__v2df) __B,
+          (__v2df) __W,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask_sqrt_round_sd( __W, __U, __A, __B, __R) __extension__ ({\
+__builtin_ia32_sqrtsd_round_mask ( (__v2df) __A,\
+          (__v2df) __B,\
+          (__v2df) __W,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128d __DEFAULT_FN_ATTRS
+_mm_maskz_sqrt_sd (__mmask8 __U, __m128d __A, __m128d __B) 
+{
+ return (__m128d) __builtin_ia32_sqrtsd_round_mask ( (__v2df) __A,
+          (__v2df) __B,
+          (__v2df) _mm_setzero_pd (),
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_maskz_sqrt_round_sd( __U, __A, __B, __R) __extension__ ({\
+__builtin_ia32_sqrtsd_round_mask ( (__v2df) __A,\
+          (__v2df) __B,\
+          (__v2df) _mm_setzero_pd (),\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask_fmadd_ss (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B) 
+{
+ return (__m128) __builtin_ia32_vfmaddss3_mask ((__v4sf) __A,
+          (__v4sf) __B,
+          (__v4sf) __W,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask_fmadd_round_ss( __W, __U, __A, __B, __R) __extension__({\
+__builtin_ia32_vfmaddss3_mask ((__v4sf) __A,\
+          (__v4sf) __B,\
+          (__v4sf) __W,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_maskz_fmadd_ss (__mmask8 __U, __m128 __A, __m128 __B, __m128 __C) 
+{
+ return (__m128) __builtin_ia32_vfmaddss3_maskz ((__v4sf) __A,
+          (__v4sf) __B,
+          (__v4sf) __C,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_maskz_fmadd_round_ss( __U, __A, __B, __C, __R) __extension__ ({\
+__builtin_ia32_vfmaddss3_maskz ((__v4sf) __A,\
+          (__v4sf) __B,\
+          (__v4sf) __C,\
+          (__mmask8) __U,\
+          _MM_FROUND_CUR_DIRECTION);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask3_fmadd_ss (__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U) 
+{
+ return (__m128) __builtin_ia32_vfmaddss3_mask3 ((__v4sf) __W,
+          (__v4sf) __X,
+          (__v4sf) __Y,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask3_fmadd_round_ss( __W, __X, __Y, __U, __R) __extension__ ({\
+__builtin_ia32_vfmaddss3_mask3 ((__v4sf) __W,\
+          (__v4sf) __X,\
+          (__v4sf) __Y,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask_fmsub_ss (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B) 
+{
+ return (__m128) __builtin_ia32_vfmaddss3_mask ((__v4sf) __A,
+          (__v4sf) (-(__B)),
+          (__v4sf) __W,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask_fmsub_round_ss( __W, __U, __A, __B, __R) __extension__ ({\
+__builtin_ia32_vfmaddss3_mask ((__v4sf) __A,\
+          (__v4sf) (-(__B)),\
+          (__v4sf) __W,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_maskz_fmsub_ss (__mmask8 __U, __m128 __A, __m128 __B, __m128 __C) 
+{
+ return (__m128) __builtin_ia32_vfmaddss3_maskz ((__v4sf) __A,
+          (__v4sf) __B,
+          (__v4sf) __C,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_maskz_fmsub_round_ss( __U, __A, __B, __C, __R) __extension__ ({\
+__builtin_ia32_vfmaddss3_maskz ((__v4sf) __A,\
+          (__v4sf) __B,\
+          (__v4sf) __C,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask3_fmsub_ss (__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U) 
+{
+ return (__m128) __builtin_ia32_vfmaddss3_mask3 ((__v4sf) __W,
+          (__v4sf) __X,
+          (__v4sf) (-__Y),
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask3_fmsub_round_ss( __W, __X, __Y, __U, __R) __extension__ ({\
+__builtin_ia32_vfmaddss3_mask3 ((__v4sf) __W,\
+          (__v4sf) __X,\
+          (__v4sf) - (__Y),\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask_fnmadd_ss (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B) 
+{
+ return (__m128) __builtin_ia32_vfmaddss3_mask ((__v4sf) -(__A),
+          (__v4sf) __B,
+          (__v4sf) __W,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask_fnmadd_round_ss( __W, __U, __A, __B, __R) __extension__ ({\
+__builtin_ia32_vfmaddss3_mask ((__v4sf) -(__A),\
+          (__v4sf) __B,\
+          (__v4sf) __W,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_maskz_fnmadd_ss (__mmask8 __U, __m128 __A, __m128 __B, __m128 __C) 
+{
+ return (__m128) __builtin_ia32_vfmaddss3_maskz ((__v4sf) -__A,
+          (__v4sf) __B,
+          (__v4sf) __C,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_maskz_fnmadd_round_ss( __U, __A, __B, __C, __R) __extension__ ({\
+__builtin_ia32_vfmaddss3_maskz ((__v4sf) -(__A),\
+          (__v4sf) __B,\
+          (__v4sf) __C,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask3_fnmadd_ss (__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U) 
+{
+ return (__m128) __builtin_ia32_vfmaddss3_mask3 ((__v4sf) -(__W),
+          (__v4sf) __X,
+          (__v4sf) __Y,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask3_fnmadd_round_ss( __W,  __X, __Y, __U, __R) __extension__({\
+__builtin_ia32_vfmaddss3_mask3 ((__v4sf) -(__W),\
+          (__v4sf) __X,\
+          (__v4sf) __Y,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask_fnmsub_ss (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B) 
+{
+ return (__m128) __builtin_ia32_vfmaddss3_mask ((__v4sf) -(__A),
+          (__v4sf) __B,
+          (__v4sf) -(__W),
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask_fnmsub_round_ss( __W, __U, __A, __B, __R) __extension__ ({\
+__builtin_ia32_vfmaddss3_mask ((__v4sf) -(__A),\
+          (__v4sf) __B,\
+          (__v4sf) -(__W),\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_maskz_fnmsub_ss (__mmask8 __U, __m128 __A, __m128 __B, __m128 __C) 
+{
+ return (__m128) __builtin_ia32_vfmaddss3_maskz ((__v4sf) -(__A),
+          (__v4sf) __B,
+          (__v4sf) -(__C),
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_maskz_fnmsub_round_ss( __U, __A, __B, __C, __R) __extension__ ({\
+__builtin_ia32_vfmaddss3_maskz((__v4sf) -(__A),\
+          (__v4sf) __B,\
+          (__v4sf) -(__C),\
+          (__mmask8) __U,\
+          _MM_FROUND_CUR_DIRECTION);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask3_fnmsub_ss (__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U) 
+{
+ return (__m128) __builtin_ia32_vfmaddss3_mask3 ((__v4sf) __W,
+          (__v4sf) __X,
+          (__v4sf) __Y,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask3_fnmsub_round_ss( __W, __X, __Y, __U, __R) __extension__({\
+__builtin_ia32_vfmaddss3_mask3 ((__v4sf) __W,\
+          (__v4sf) __X,\
+          (__v4sf) __Y,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask_fmadd_sd (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B) 
+{
+ return (__m128) __builtin_ia32_vfmaddsd3_mask ( (__v2df) __A,
+          (__v2df) __B,
+          (__v2df) __W,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask_fmadd_round_sd( __W, __U, __A, __B, __R) __extension__({\
+__builtin_ia32_vfmaddsd3_mask ( (__v2df) __A,\
+          (__v2df) __B,\
+          (__v2df) __W,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_maskz_fmadd_sd (__mmask8 __U, __m128 __A, __m128 __B, __m128 __C) 
+{
+ return (__m128) __builtin_ia32_vfmaddsd3_maskz ( (__v2df) __A,
+          (__v2df) __B,
+          (__v2df) __C,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_maskz_fmadd_round_sd( __U, __A, __B, __C, __R) __extension__ ({\
+__builtin_ia32_vfmaddsd3_maskz ( (__v2df) __A,\
+          (__v2df) __B,\
+          (__v2df) __C,\
+          (__mmask8) __U,\
+          _MM_FROUND_CUR_DIRECTION);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask3_fmadd_sd (__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U) 
+{
+ return (__m128) __builtin_ia32_vfmaddsd3_mask3 ((__v2df) __W,
+          (__v2df) __X,
+          (__v2df) __Y,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask3_fmadd_round_sd( __W, __X, __Y, __U, __R) __extension__ ({\
+__builtin_ia32_vfmaddsd3_mask3 ((__v2df) __W,\
+          (__v2df) __X,\
+          (__v2df) __Y,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask_fmsub_sd (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B) 
+{
+ return (__m128) __builtin_ia32_vfmaddsd3_mask ( (__v2df) __A,
+          (__v2df) (-(__B)),
+          (__v2df) __W,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask_fmsub_round_sd( __W, __U, __A, __B, __R) __extension__ ({\
+__builtin_ia32_vfmaddsd3_mask ( (__v2df) __A,\
+          (__v2df) (-(__B)),\
+          (__v2df) __W,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_maskz_fmsub_sd (__mmask8 __U, __m128 __A, __m128 __B, __m128 __C) 
+{
+ return (__m128) __builtin_ia32_vfmaddsd3_maskz ( (__v2df) __A,
+          (__v2df) __B,
+          (__v2df) __C,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_maskz_fmsub_round_sd( __U, __A, __B, __C, __R) __extension__ ({\
+__builtin_ia32_vfmaddsd3_maskz ( (__v2df) __A,\
+          (__v2df) __B,\
+          (__v2df) __C,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask3_fmsub_sd (__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U) 
+{
+ return (__m128) __builtin_ia32_vfmaddsd3_mask3 ((__v2df) __W,
+          (__v2df) __X,
+          (__v2df) (-__Y),
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask3_fmsub_round_sd( __W, __X, __Y, __U, __R) __extension__ ({\
+__builtin_ia32_vfmaddsd3_mask3 ((__v2df) __W,\
+          (__v2df) __X,\
+          (__v2df) -(__Y),\
+          (__mmask8) __U, __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask_fnmadd_sd (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B) 
+{
+ return (__m128) __builtin_ia32_vfmaddsd3_mask ( (__v2df) -(__A),
+          (__v2df) __B,
+          (__v2df) __W,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask_fnmadd_round_sd( __W, __U, __A, __B, __R) __extension__ ({\
+__builtin_ia32_vfmaddsd3_mask ( (__v2df) -(__A),\
+          (__v2df) __B,\
+          (__v2df) __W,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_maskz_fnmadd_sd (__mmask8 __U, __m128 __A, __m128 __B, __m128 __C) 
+{
+ return (__m128) __builtin_ia32_vfmaddsd3_maskz ( (__v2df) -__A,
+          (__v2df) __B,
+          (__v2df) __C,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_maskz_fnmadd_round_sd( __U, __A, __B, __C, __R) __extension__ ({\
+__builtin_ia32_vfmaddsd3_maskz ( (__v2df) -(__A),\
+          (__v2df) __B,\
+          (__v2df) __C,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask3_fnmadd_sd (__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U) 
+{
+ return (__m128) __builtin_ia32_vfmaddsd3_mask3 ((__v2df) -(__W),
+          (__v2df) __X,
+          (__v2df) __Y,
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask3_fnmadd_round_sd( __W,  __X, __Y, __U, __R) __extension__({\
+__builtin_ia32_vfmaddsd3_mask3 ((__v2df) -(__W),\
+          (__v2df) __X,\
+          (__v2df) __Y,\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask_fnmsub_sd (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B) 
+{
+ return (__m128) __builtin_ia32_vfmaddsd3_mask ( (__v2df) -(__A),
+          (__v2df) __B,
+          (__v2df) -(__W),
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask_fnmsub_round_sd( __W, __U, __A, __B, __R) __extension__ ({\
+__builtin_ia32_vfmaddsd3_mask ( (__v2df) -(__A),\
+          (__v2df) __B,\
+          (__v2df) -(__W),\
+          (__mmask8) __U,\
+          __R);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_maskz_fnmsub_sd (__mmask8 __U, __m128 __A, __m128 __B, __m128 __C) 
+{
+ return (__m128) __builtin_ia32_vfmaddsd3_maskz ( (__v2df) -(__A),
+          (__v2df) __B,
+          (__v2df) -(__C),
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_maskz_fnmsub_round_sd( __U, __A, __B, __C, __R) __extension__ ({\
+__builtin_ia32_vfmaddsd3_maskz( (__v2df) -(__A),\
+          (__v2df) __B,\
+          (__v2df) -(__C),\
+          (__mmask8) __U,\
+          _MM_FROUND_CUR_DIRECTION);\
+})
+
+static __inline__ __m128 __DEFAULT_FN_ATTRS
+_mm_mask3_fnmsub_sd (__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U) 
+{
+ return (__m128) __builtin_ia32_vfmaddsd3_mask3 ((__v2df) -(__W),
+          (__v2df) __X,
+          (__v2df) -(__Y),
+          (__mmask8) __U,
+          _MM_FROUND_CUR_DIRECTION);
+}
+
+#define _mm_mask3_fnmsub_round_sd( __W, __X, __Y, __U, __R) __extension__({\
+__builtin_ia32_vfmaddsd3_mask3 ((__v2df) -(__W),\
+          (__v2df) __X,\
+          (__v2df) -(__Y),\
+          (__mmask8) __U,\
+          __R);\
+})
 
 #undef __DEFAULT_FN_ATTRS
 
 #endif // __AVX512FINTRIN_H

test/CodeGen/avx512f-builtins.c

 }
 
 __mmask16 test_mm512_kmov(__mmask16 __A) {
   // CHECK-LABEL: @test_mm512_kmov
   // CHECK: load i16, i16* %__A.addr.i, align 2
   return _mm512_kmov(__A); 
 }
 
+__m128 test_mm_sqrt_round_ss(__m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_sqrt_round_ss
+  // CHECK: @llvm.x86.avx512.mask.sqrt.ss
+  return _mm_sqrt_round_ss(__A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_mask_sqrt_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_sqrt_ss
+  // CHECK: @llvm.x86.avx512.mask.sqrt.ss
+  return _mm_mask_sqrt_ss(__W, __U, __A, __B);
+}
+
+__m128 test_mm_mask_sqrt_round_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_sqrt_round_ss
+  // CHECK: @llvm.x86.avx512.mask.sqrt.ss
+  return _mm_mask_sqrt_round_ss(__W, __U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_maskz_sqrt_ss(__mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_maskz_sqrt_ss
+  // CHECK: @llvm.x86.avx512.mask.sqrt.ss
+  return _mm_maskz_sqrt_ss(__U, __A, __B);
+}
+
+__m128 test_mm_maskz_sqrt_round_ss(__mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_maskz_sqrt_round_ss
+  // CHECK: @llvm.x86.avx512.mask.sqrt.ss
+  return _mm_maskz_sqrt_round_ss(__U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
 
+__m128d test_mm_sqrt_round_sd(__m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_sqrt_round_sd
+  // CHECK: @llvm.x86.avx512.mask.sqrt.sd
+  return _mm_sqrt_round_sd(__A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_mask_sqrt_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_sqrt_sd
+  // CHECK: @llvm.x86.avx512.mask.sqrt.sd
+  return _mm_mask_sqrt_sd(__W, __U, __A, __B);
+}
 
+__m128d test_mm_mask_sqrt_round_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_sqrt_round_sd
+  // CHECK: @llvm.x86.avx512.mask.sqrt.sd
+  return _mm_mask_sqrt_round_sd(__W, __U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_maskz_sqrt_sd(__mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_maskz_sqrt_sd
+  // CHECK: @llvm.x86.avx512.mask.sqrt.sd
+  return _mm_maskz_sqrt_sd(__U, __A, __B);
+}
+
+__m128d test_mm_maskz_sqrt_round_sd(__mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_maskz_sqrt_round_sd
+  // CHECK: @llvm.x86.avx512.mask.sqrt.sd
+  return _mm_maskz_sqrt_round_sd(__U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_mask_fmadd_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_fmadd_ss
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.ss
+  return _mm_mask_fmadd_ss(__W, __U, __A, __B);
+}
+
+__m128 test_mm_mask_fmadd_round_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_fmadd_round_ss
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.ss
+  return _mm_mask_fmadd_round_ss(__W, __U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_maskz_fmadd_ss(__mmask8 __U, __m128 __A, __m128 __B, __m128 __C){
+  // CHECK-LABEL: @test_mm_maskz_fmadd_ss
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.ss
+  return _mm_maskz_fmadd_ss(__U, __A, __B, __C);
+}
+
+__m128 test_mm_maskz_fmadd_round_ss(__mmask8 __U, __m128 __A, __m128 __B, __m128 __C){
+  // CHECK-LABEL: @test_mm_maskz_fmadd_round_ss
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.ss
+  return _mm_maskz_fmadd_round_ss(__U, __A, __B, __C, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_mask3_fmadd_ss(__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fmadd_ss
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.ss
+  return _mm_mask3_fmadd_ss(__W, __X, __Y, __U);
+}
+
+__m128 test_mm_mask3_fmadd_round_ss(__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fmadd_round_ss
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.ss
+  return _mm_mask3_fmadd_round_ss(__W, __X, __Y, __U, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_mask_fmsub_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_fmsub_ss
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.ss
+  return _mm_mask_fmsub_ss(__W, __U, __A, __B);
+}
+
+__m128 test_mm_mask_fmsub_round_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_fmsub_round_ss
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.ss
+  return _mm_mask_fmsub_round_ss(__W, __U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_maskz_fmsub_ss(__mmask8 __U, __m128 __A, __m128 __B, __m128 __C){
+  // CHECK-LABEL: @test_mm_maskz_fmsub_ss
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.ss
+  return _mm_maskz_fmsub_ss(__U, __A, __B, __C);
+}
+
+__m128 test_mm_maskz_fmsub_round_ss(__mmask8 __U, __m128 __A, __m128 __B, __m128 __C){
+  // CHECK-LABEL: @test_mm_maskz_fmsub_round_ss
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.ss
+  return _mm_maskz_fmsub_round_ss(__U, __A, __B, __C, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_mask3_fmsub_ss(__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fmsub_ss
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.ss
+  return _mm_mask3_fmsub_ss(__W, __X, __Y, __U);
+}
+
+__m128 test_mm_mask3_fmsub_round_ss(__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fmsub_round_ss
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.ss
+  return _mm_mask3_fmsub_round_ss(__W, __X, __Y, __U, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_mask_fnmadd_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_fnmadd_ss
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.ss
+  return _mm_mask_fnmadd_ss(__W, __U, __A, __B);
+}
+
+__m128 test_mm_mask_fnmadd_round_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_fnmadd_round_ss
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.ss
+  return _mm_mask_fnmadd_round_ss(__W, __U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_maskz_fnmadd_ss(__mmask8 __U, __m128 __A, __m128 __B, __m128 __C){
+  // CHECK-LABEL: @test_mm_maskz_fnmadd_ss
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.ss
+  return _mm_maskz_fnmadd_ss(__U, __A, __B, __C);
+}
+
+__m128 test_mm_maskz_fnmadd_round_ss(__mmask8 __U, __m128 __A, __m128 __B, __m128 __C){
+  // CHECK-LABEL: @test_mm_maskz_fnmadd_round_ss
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.ss
+  return _mm_maskz_fnmadd_round_ss(__U, __A, __B, __C, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_mask3_fnmadd_ss(__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fnmadd_ss
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.ss
+  return _mm_mask3_fnmadd_ss(__W, __X, __Y, __U);
+}
+
+__m128 test_mm_mask3_fnmadd_round_ss(__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fnmadd_round_ss
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.ss
+  return _mm_mask3_fnmadd_round_ss(__W, __X, __Y, __U, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_mask_fnmsub_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_fnmsub_ss
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.ss
+  return _mm_mask_fnmsub_ss(__W, __U, __A, __B);
+}
+
+__m128 test_mm_mask_fnmsub_round_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_fnmsub_round_ss
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.ss
+  return _mm_mask_fnmsub_round_ss(__W, __U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_maskz_fnmsub_ss(__mmask8 __U, __m128 __A, __m128 __B, __m128 __C){
+  // CHECK-LABEL: @test_mm_maskz_fnmsub_ss
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.ss
+  return _mm_maskz_fnmsub_ss(__U, __A, __B, __C);
+}
+
+__m128 test_mm_maskz_fnmsub_round_ss(__mmask8 __U, __m128 __A, __m128 __B, __m128 __C){
+  // CHECK-LABEL: @test_mm_maskz_fnmsub_round_ss
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.ss
+  return _mm_maskz_fnmsub_round_ss(__U, __A, __B, __C, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_mask3_fnmsub_ss(__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fnmsub_ss
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.ss
+  return _mm_mask3_fnmsub_ss(__W, __X, __Y, __U);
+}
+
+__m128 test_mm_mask3_fnmsub_round_ss(__m128 __W, __m128 __X, __m128 __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fnmsub_round_ss
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.ss
+  return _mm_mask3_fnmsub_round_ss(__W, __X, __Y, __U, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_mask_fmadd_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_fmadd_sd
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.sd
+  return _mm_mask_fmadd_sd(__W, __U, __A, __B);
+}
+
+__m128d test_mm_mask_fmadd_round_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_fmadd_round_sd
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.sd
+  return _mm_mask_fmadd_round_sd(__W, __U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_maskz_fmadd_sd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C){
+  // CHECK-LABEL: @test_mm_maskz_fmadd_sd
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.sd
+  return _mm_maskz_fmadd_sd(__U, __A, __B, __C);
+}
+
+__m128d test_mm_maskz_fmadd_round_sd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C){
+  // CHECK-LABEL: @test_mm_maskz_fmadd_round_sd
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.sd
+  return _mm_maskz_fmadd_round_sd(__U, __A, __B, __C, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_mask3_fmadd_sd(__m128d __W, __m128d __X, __m128d __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fmadd_sd
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.sd
+  return _mm_mask3_fmadd_sd(__W, __X, __Y, __U);
+}
+
+__m128d test_mm_mask3_fmadd_round_sd(__m128d __W, __m128d __X, __m128d __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fmadd_round_sd
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.sd
+  return _mm_mask3_fmadd_round_sd(__W, __X, __Y, __U, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_mask_fmsub_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_fmsub_sd
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.sd
+  return _mm_mask_fmsub_sd(__W, __U, __A, __B);
+}
+
+__m128d test_mm_mask_fmsub_round_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_fmsub_round_sd
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.sd
+  return _mm_mask_fmsub_round_sd(__W, __U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_maskz_fmsub_sd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C){
+  // CHECK-LABEL: @test_mm_maskz_fmsub_sd
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.sd
+  return _mm_maskz_fmsub_sd(__U, __A, __B, __C);
+}
+
+__m128d test_mm_maskz_fmsub_round_sd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C){
+  // CHECK-LABEL: @test_mm_maskz_fmsub_round_sd
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.sd
+  return _mm_maskz_fmsub_round_sd(__U, __A, __B, __C, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_mask3_fmsub_sd(__m128d __W, __m128d __X, __m128d __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fmsub_sd
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.sd
+  return _mm_mask3_fmsub_sd(__W, __X, __Y, __U);
+}
+
+__m128d test_mm_mask3_fmsub_round_sd(__m128d __W, __m128d __X, __m128d __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fmsub_round_sd
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.sd
+  return _mm_mask3_fmsub_round_sd(__W, __X, __Y, __U, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_mask_fnmadd_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_fnmadd_sd
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.sd
+  return _mm_mask_fnmadd_sd(__W, __U, __A, __B);
+}
+
+__m128d test_mm_mask_fnmadd_round_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_fnmadd_round_sd
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.sd
+  return _mm_mask_fnmadd_round_sd(__W, __U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_maskz_fnmadd_sd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C){
+  // CHECK-LABEL: @test_mm_maskz_fnmadd_sd
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.sd
+  return _mm_maskz_fnmadd_sd(__U, __A, __B, __C);
+}
+
+__m128d test_mm_maskz_fnmadd_round_sd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C){
+  // CHECK-LABEL: @test_mm_maskz_fnmadd_round_sd
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.sd
+  return _mm_maskz_fnmadd_round_sd(__U, __A, __B, __C, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_mask3_fnmadd_sd(__m128d __W, __m128d __X, __m128d __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fnmadd_sd
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.sd
+  return _mm_mask3_fnmadd_sd(__W, __X, __Y, __U);
+}
+
+__m128d test_mm_mask3_fnmadd_round_sd(__m128d __W, __m128d __X, __m128d __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fnmadd_round_sd
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.sd
+  return _mm_mask3_fnmadd_round_sd(__W, __X, __Y, __U, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_mask_fnmsub_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_fnmsub_sd
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.sd
+  return _mm_mask_fnmsub_sd(__W, __U, __A, __B);
+}
+
+__m128d test_mm_mask_fnmsub_round_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_fnmsub_round_sd
+  // CHECK: @llvm.x86.avx512.mask.vfmadd.sd
+  return _mm_mask_fnmsub_round_sd(__W, __U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_maskz_fnmsub_sd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C){
+  // CHECK-LABEL: @test_mm_maskz_fnmsub_sd
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.sd
+  return _mm_maskz_fnmsub_sd(__U, __A, __B, __C);
+}
+
+__m128d test_mm_maskz_fnmsub_round_sd(__mmask8 __U, __m128d __A, __m128d __B, __m128d __C){
+  // CHECK-LABEL: @test_mm_maskz_fnmsub_round_sd
+  // CHECK: @llvm.x86.avx512.maskz.vfmadd.sd
+  return _mm_maskz_fnmsub_round_sd(__U, __A, __B, __C, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_mask3_fnmsub_sd(__m128d __W, __m128d __X, __m128d __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fnmsub_sd
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.sd
+  return _mm_mask3_fnmsub_sd(__W, __X, __Y, __U);
+}
+
+__m128d test_mm_mask3_fnmsub_round_sd(__m128d __W, __m128d __X, __m128d __Y, __mmask8 __U){
+  // CHECK-LABEL: @test_mm_mask3_fnmsub_round_sd
+  // CHECK: @llvm.x86.avx512.mask3.vfmadd.sd
+  return _mm_mask3_fnmsub_round_sd(__W, __X, __Y, __U, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_mask_rsqrt14_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_rsqrt14_sd
+  // CHECK: @llvm.x86.avx512.rsqrt14.sd
+  return _mm_mask_rsqrt14_sd(__W, __U, __A, __B);
+}
+
+__m128d test_mm_maskz_rsqrt14_sd(__mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_maskz_rsqrt14_sd
+  // CHECK: @llvm.x86.avx512.rsqrt14.sd
+  return _mm_maskz_rsqrt14_sd(__U, __A, __B);
+}
+
+__m128 test_mm_mask_rsqrt14_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_rsqrt14_ss
+  // CHECK: @llvm.x86.avx512.rsqrt14.ss
+  return _mm_mask_rsqrt14_ss(__W, __U, __A, __B);
+}
+
+__m128 test_mm_maskz_rsqrt14_ss(__mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_maskz_rsqrt14_ss
+  // CHECK: @llvm.x86.avx512.rsqrt14.ss
+  return _mm_maskz_rsqrt14_ss(__U, __A, __B);
+}
+
+__m128d test_mm_mask_rcp14_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_rcp14_sd
+  // CHECK: @llvm.x86.avx512.rcp14.sd
+  return _mm_mask_rcp14_sd(__W, __U, __A, __B);
+}
+
+__m128d test_mm_maskz_rcp14_sd(__mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_maskz_rcp14_sd
+  // CHECK: @llvm.x86.avx512.rcp14.sd
+  return _mm_maskz_rcp14_sd(__U, __A, __B);
+}
+
+__m128 test_mm_mask_rcp14_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_rcp14_ss
+  // CHECK: @llvm.x86.avx512.rcp14.ss
+  return _mm_mask_rcp14_ss(__W, __U, __A, __B);
+}
+
+__m128 test_mm_maskz_rcp14_ss(__mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_maskz_rcp14_ss
+  // CHECK: @llvm.x86.avx512.rcp14.ss
+  return _mm_maskz_rcp14_ss(__U, __A, __B);
+}
+
+__m128d test_mm_mask_getexp_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_getexp_sd
+  // CHECK: @llvm.x86.avx512.mask.getexp.sd
+  return _mm_mask_getexp_sd(__W, __U, __A, __B);
+}
+
+__m128d test_mm_mask_getexp_round_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_getexp_round_sd
+  // CHECK: @llvm.x86.avx512.mask.getexp.sd
+  return _mm_mask_getexp_round_sd(__W, __U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_maskz_getexp_sd(__mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_maskz_getexp_sd
+  // CHECK: @llvm.x86.avx512.mask.getexp.sd
+  return _mm_maskz_getexp_sd(__U, __A, __B);
+}
+
+__m128d test_mm_maskz_getexp_round_sd(__mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_maskz_getexp_round_sd
+  // CHECK: @llvm.x86.avx512.mask.getexp.sd
+  return _mm_maskz_getexp_round_sd(__U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_mask_getexp_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_getexp_ss
+  // CHECK: @llvm.x86.avx512.mask.getexp.ss
+  return _mm_mask_getexp_ss(__W, __U, __A, __B);
+}
+
+__m128 test_mm_mask_getexp_round_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_getexp_round_ss
+  // CHECK: @llvm.x86.avx512.mask.getexp.ss
+  return _mm_mask_getexp_round_ss(__W, __U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_maskz_getexp_ss(__mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_maskz_getexp_ss
+  // CHECK: @llvm.x86.avx512.mask.getexp.ss
+  return _mm_maskz_getexp_ss(__U, __A, __B);
+}
+
+__m128 test_mm_maskz_getexp_round_ss(__mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_maskz_getexp_round_ss
+  // CHECK: @llvm.x86.avx512.mask.getexp.ss
+  return _mm_maskz_getexp_round_ss(__U, __A, __B, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_mask_getmant_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_getmant_sd
+  // CHECK: @llvm.x86.avx512.mask.getmant.sd
+  return _mm_mask_getmant_sd(__W, __U, __A, __B, 1, 2);
+}
+
+__m128d test_mm_mask_getmant_round_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_mask_getmant_round_sd
+  // CHECK: @llvm.x86.avx512.mask.getmant.sd
+  return _mm_mask_getmant_round_sd(__W, __U, __A, __B, 1, 2, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128d test_mm_maskz_getmant_sd(__mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_maskz_getmant_sd
+  // CHECK: @llvm.x86.avx512.mask.getmant.sd
+  return _mm_maskz_getmant_sd(__U, __A, __B, 1, 2);
+}
+
+__m128d test_mm_maskz_getmant_round_sd(__mmask8 __U, __m128d __A, __m128d __B){
+  // CHECK-LABEL: @test_mm_maskz_getmant_round_sd
+  // CHECK: @llvm.x86.avx512.mask.getmant.sd
+  return _mm_maskz_getmant_round_sd(__U, __A, __B, 1, 2, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_mask_getmant_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_getmant_ss
+  // CHECK: @llvm.x86.avx512.mask.getmant.ss
+  return _mm_mask_getmant_ss(__W, __U, __A, __B, 1, 2);
+}
+
+__m128 test_mm_mask_getmant_round_ss(__m128 __W, __mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_mask_getmant_round_ss
+  // CHECK: @llvm.x86.avx512.mask.getmant.ss
+  return _mm_mask_getmant_round_ss(__W, __U, __A, __B, 1, 2, _MM_FROUND_CUR_DIRECTION);
+}
+
+__m128 test_mm_maskz_getmant_ss(__mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_maskz_getmant_ss
+  // CHECK: @llvm.x86.avx512.mask.getmant.ss
+  return _mm_maskz_getmant_ss(__U, __A, __B, 1, 2);
+}
+
+__m128 test_mm_maskz_getmant_round_ss(__mmask8 __U, __m128 __A, __m128 __B){
+  // CHECK-LABEL: @test_mm_maskz_getmant_round_ss
+  // CHECK: @llvm.x86.avx512.mask.getmant.ss
+  return _mm_maskz_getmant_round_ss(__U, __A, __B, 1, 2, _MM_FROUND_CUR_DIRECTION);
+}