Add missing REDUCE[SD,SS] intrinsics

gcc/
	* config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
	_mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
	_mm_maskz_reduce_ss): New.
	* config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
	__builtin_ia32_reducess_mask): Ditto..
	(__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
	* config/i386/sse.md (reduces<mode>): Renamed to ...
	(reduces<mode><mask_scalar_name>): ... this.
	(vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
	to ...
	(vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
	%0<mask_scalar_operand4>, %1, %2, %3}): ... this.

gcc/testsuite/
	* gcc.target/i386/avx512dq-vreducesd-1.c (_mm_mask_reduce_sd,
	_mm_maskz_reduce_sd): Test new intrinsics.
	* gcc.target/i386/avx512dq-vreducesd-2.c: New.
	* gcc.target/i386/avx512dq-vreducess-1.c (_mm_mask_reduce_ss,
	_mm_maskz_reduce_ss): Test new intrinsics.
	* gcc.target/i386/avx512dq-vreducess-2.c: New.
	* gcc.target/i386/avx-1.c (__builtin_ia32_reducesd,
	__builtin_ia32_reducess): Remove builtin.
	(__builtin_ia32_reducesd_mask,
	__builtin_ia32_reducess_mask): Test new builtin.
	* gcc.target/i386/sse-13.c: Ditto.
	* gcc.target/i386/sse-23.c: Ditto.

From-SVN: r253803

gcc/ChangeLog

+17-10-2017  Olga Makhotina  <olga.makhotina@intel.com>
+
+	* config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
+	_mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
+	_mm_maskz_reduce_ss): New.
+	* config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
+	__builtin_ia32_reducess_mask): Ditto..
+	(__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
+	* config/i386/sse.md (reduces<mode>): Renamed to ...
+	(reduces<mode><mask_scalar_name>): ... this.
+	(vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
+	to ...
+	(vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
+	%0<mask_scalar_operand4>, %1, %2, %3}): ... this.
+
 2017-10-16  David Malcolm  <dmalcolm@redhat.com>
 
 	* Makefile.in (OBJS): Add unique-ptr-tests.o.

gcc/config/i386/avx512dqintrin.h

@@ -1160,16 +1160,63 @@ extern __inline __m128d
 __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
 _mm_reduce_sd (__m128d __A, __m128d __B, int __C)
 {
-  return (__m128d) __builtin_ia32_reducesd ((__v2df) __A,
-						 (__v2df) __B, __C);
+  return (__m128d) __builtin_ia32_reducesd_mask ((__v2df) __A,
+						 (__v2df) __B, __C,
+						 (__v2df) _mm_setzero_pd (),
+						 (__mmask8) -1);
+}
+
+extern __inline __m128d
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_reduce_sd (__m128d __W,  __mmask8 __U, __m128d __A,
+		    __m128d __B, int __C)
+{
+  return (__m128d) __builtin_ia32_reducesd_mask ((__v2df) __A,
+						 (__v2df) __B, __C,
+						 (__v2df) __W,
+						 (__mmask8) __U);
+}
+
+extern __inline __m128d
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_maskz_reduce_sd (__mmask8 __U, __m128d __A, __m128d __B, int __C)
+{
+  return (__m128d) __builtin_ia32_reducesd_mask ((__v2df) __A,
+						 (__v2df) __B, __C,
+						 (__v2df) _mm_setzero_pd (),
+						 (__mmask8) __U);
 }
 
 extern __inline __m128
 __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
 _mm_reduce_ss (__m128 __A, __m128 __B, int __C)
 {
-  return (__m128) __builtin_ia32_reducess ((__v4sf) __A,
-						(__v4sf) __B, __C);
+  return (__m128) __builtin_ia32_reducess_mask ((__v4sf) __A,
+						(__v4sf) __B, __C,
+						(__v4sf) _mm_setzero_ps (),
+						(__mmask8) -1);
+}
+
+
+extern __inline __m128
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_reduce_ss (__m128 __W,  __mmask8 __U, __m128 __A,
+		    __m128 __B, int __C)
+{
+  return (__m128) __builtin_ia32_reducess_mask ((__v4sf) __A,
+						(__v4sf) __B, __C,
+						(__v4sf) __W,
+						(__mmask8) __U);
+}
+
+extern __inline __m128
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_maskz_reduce_ss (__mmask8 __U, __m128 __A, __m128 __B, int __C)
+{
+  return (__m128) __builtin_ia32_reducess_mask ((__v4sf) __A,
+						(__v4sf) __B, __C,
+						(__v4sf) _mm_setzero_ps (),
+						(__mmask8) __U);
 }
 
 extern __inline __m128d
@@ -2449,12 +2496,34 @@ _mm512_fpclass_ps_mask (__m512 __A, const int __imm)
 						 (int) (c),(__mmask8)-1))
 
 #define _mm_reduce_sd(A, B, C)						\
-  ((__m128d) __builtin_ia32_reducesd ((__v2df)(__m128d)(A),	\
-    (__v2df)(__m128d)(B), (int)(C)))					\
+  ((__m128d) __builtin_ia32_reducesd_mask ((__v2df)(__m128d)(A),	\
+    (__v2df)(__m128d)(B), (int)(C), (__v2df) _mm_setzero_pd (),		\
+    (__mmask8)-1))
+
+#define _mm_mask_reduce_sd(W, U, A, B, C)				\
+  ((__m128d) __builtin_ia32_reducesd_mask ((__v2df)(__m128d)(A),	\
+    (__v2df)(__m128d)(B), (int)(C), (__v2df)(__m128d)(W), (__mmask8)(U)))
+
+#define _mm_maskz_reduce_sd(U, A, B, C)					\
+  ((__m128d) __builtin_ia32_reducesd_mask ((__v2df)(__m128d)(A),	\
+    (__v2df)(__m128d)(B), (int)(C), (__v2df) _mm_setzero_pd (),		\
+    (__mmask8)(U)))
 
 #define _mm_reduce_ss(A, B, C)						\
-  ((__m128) __builtin_ia32_reducess ((__v4sf)(__m128)(A),		\
-    (__v4sf)(__m128)(A), (int)(C)))					\
+  ((__m128) __builtin_ia32_reducess_mask ((__v4sf)(__m128)(A),		\
+    (__v4sf)(__m128)(B), (int)(C), (__v4sf) _mm_setzero_ps (),		\
+    (__mmask8)-1))
+
+#define _mm_mask_reduce_ss(W, U, A, B, C)				\
+  ((__m128) __builtin_ia32_reducess_mask ((__v4sf)(__m128)(A),		\
+    (__v4sf)(__m128)(B), (int)(C), (__v4sf)(__m128)(W), (__mmask8)(U)))
+
+#define _mm_maskz_reduce_ss(U, A, B, C)					\
+  ((__m128) __builtin_ia32_reducess_mask ((__v4sf)(__m128)(A),		\
+    (__v4sf)(__m128)(B), (int)(C), (__v4sf) _mm_setzero_ps (),		\
+    (__mmask8)(U)))
+
+
 
 #endif
 

gcc/config/i386/i386-builtin.def

@@ -1666,8 +1666,8 @@ BDESC (OPTION_MASK_ISA_AVX512DQ | OPTION_MASK_ISA_AVX512VL, CODE_FOR_reducepv4df
 BDESC (OPTION_MASK_ISA_AVX512DQ | OPTION_MASK_ISA_AVX512VL, CODE_FOR_reducepv2df_mask, "__builtin_ia32_reducepd128_mask", IX86_BUILTIN_REDUCEPD128_MASK, UNKNOWN, (int) V2DF_FTYPE_V2DF_INT_V2DF_UQI)
 BDESC (OPTION_MASK_ISA_AVX512DQ | OPTION_MASK_ISA_AVX512VL, CODE_FOR_reducepv8sf_mask, "__builtin_ia32_reduceps256_mask", IX86_BUILTIN_REDUCEPS256_MASK, UNKNOWN, (int) V8SF_FTYPE_V8SF_INT_V8SF_UQI)
 BDESC (OPTION_MASK_ISA_AVX512DQ | OPTION_MASK_ISA_AVX512VL, CODE_FOR_reducepv4sf_mask, "__builtin_ia32_reduceps128_mask", IX86_BUILTIN_REDUCEPS128_MASK, UNKNOWN, (int) V4SF_FTYPE_V4SF_INT_V4SF_UQI)
-BDESC (OPTION_MASK_ISA_AVX512DQ, CODE_FOR_reducesv2df, "__builtin_ia32_reducesd", IX86_BUILTIN_REDUCESD_MASK, UNKNOWN, (int) V2DF_FTYPE_V2DF_V2DF_INT)
-BDESC (OPTION_MASK_ISA_AVX512DQ, CODE_FOR_reducesv4sf, "__builtin_ia32_reducess", IX86_BUILTIN_REDUCESS_MASK, UNKNOWN, (int) V4SF_FTYPE_V4SF_V4SF_INT)
+BDESC (OPTION_MASK_ISA_AVX512DQ, CODE_FOR_reducesv2df_mask, "__builtin_ia32_reducesd_mask", IX86_BUILTIN_REDUCESD128_MASK, UNKNOWN, (int) V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI)
+BDESC (OPTION_MASK_ISA_AVX512DQ, CODE_FOR_reducesv4sf_mask, "__builtin_ia32_reducess_mask", IX86_BUILTIN_REDUCESS128_MASK, UNKNOWN, (int) V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI)
 BDESC (OPTION_MASK_ISA_AVX512BW | OPTION_MASK_ISA_AVX512VL, CODE_FOR_avx512vl_permvarv16hi_mask, "__builtin_ia32_permvarhi256_mask", IX86_BUILTIN_VPERMVARHI256_MASK, UNKNOWN, (int) V16HI_FTYPE_V16HI_V16HI_V16HI_UHI)
 BDESC (OPTION_MASK_ISA_AVX512BW | OPTION_MASK_ISA_AVX512VL, CODE_FOR_avx512vl_permvarv8hi_mask, "__builtin_ia32_permvarhi128_mask", IX86_BUILTIN_VPERMVARHI128_MASK, UNKNOWN, (int) V8HI_FTYPE_V8HI_V8HI_V8HI_UQI)
 BDESC (OPTION_MASK_ISA_AVX512BW | OPTION_MASK_ISA_AVX512VL, CODE_FOR_avx512vl_vpermt2varv16hi3_mask, "__builtin_ia32_vpermt2varhi256_mask", IX86_BUILTIN_VPERMT2VARHI256, UNKNOWN, (int) V16HI_FTYPE_V16HI_V16HI_V16HI_UHI)

gcc/config/i386/sse.md

@@ -2522,7 +2522,7 @@
    (set_attr "prefix" "evex")
    (set_attr "mode" "<MODE>")])
 
-(define_insn "reduces<mode>"
+(define_insn "reduces<mode><mask_scalar_name>"
   [(set (match_operand:VF_128 0 "register_operand" "=v")
 	(vec_merge:VF_128
 	  (unspec:VF_128
@@ -2533,7 +2533,7 @@
 	  (match_dup 1)
 	  (const_int 1)))]
   "TARGET_AVX512DQ"
-  "vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}"
+  "vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1, %2, %3}"
   [(set_attr "type" "sse")
    (set_attr "prefix" "evex")
    (set_attr "mode" "<MODE>")])

gcc/testsuite/ChangeLog

+17-10-2017  Olga Makhotina  <olga.makhotina@intel.com>
+
+	* gcc.target/i386/avx512dq-vreducesd-1.c (_mm_mask_reduce_sd,
+	_mm_maskz_reduce_sd): Test new intrinsics.
+	* gcc.target/i386/avx512dq-vreducesd-2.c: New.
+	* gcc.target/i386/avx512dq-vreducess-1.c (_mm_mask_reduce_ss,
+	_mm_maskz_reduce_ss): Test new intrinsics.
+	* gcc.target/i386/avx512dq-vreducess-2.c: New.
+	* gcc.target/i386/avx-1.c (__builtin_ia32_reducesd,
+	__builtin_ia32_reducess): Remove builtin.
+	(__builtin_ia32_reducesd_mask,
+	__builtin_ia32_reducess_mask): Test new builtin.
+	* gcc.target/i386/sse-13.c: Ditto.
+	* gcc.target/i386/sse-23.c: Ditto.
+
 2017-10-16  Martin Liska  <mliska@suse.cz>
 
 	* c-c++-common/ubsan/attrib-5.c (float_cast2): Fix warning scan

gcc/testsuite/gcc.target/i386/avx-1.c

@@ -412,8 +412,8 @@
 /* avx512dqintrin.h */
 #define __builtin_ia32_kshiftliqi(A, B) __builtin_ia32_kshiftliqi(A, 8)
 #define __builtin_ia32_kshiftriqi(A, B) __builtin_ia32_kshiftriqi(A, 8)
-#define __builtin_ia32_reducess(A, B, F) __builtin_ia32_reducess(A, B, 1)
-#define __builtin_ia32_reducesd(A, B, F) __builtin_ia32_reducesd(A, B, 1)
+#define __builtin_ia32_reducess_mask(A, B, F, W, U) __builtin_ia32_reducess_mask(A, B, 1, W, U)
+#define __builtin_ia32_reducesd_mask(A, B, F, W, U) __builtin_ia32_reducesd_mask(A, B, 1, W, U)
 #define __builtin_ia32_reduceps512_mask(A, E, C, D) __builtin_ia32_reduceps512_mask(A, 1, C, D)
 #define __builtin_ia32_reducepd512_mask(A, E, C, D) __builtin_ia32_reducepd512_mask(A, 1, C, D)
 #define __builtin_ia32_rangess128_round(A, B, I, F) __builtin_ia32_rangess128_round(A, B, 1, 8)

gcc/testsuite/gcc.target/i386/avx512dq-vreducesd-1.c

@@ -2,13 +2,24 @@
 /* { dg-options "-mavx512dq -O2" } */
 /* { dg-final { scan-assembler-times "vreducesd\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
 
+/* { dg-final { scan-assembler-times "vreducesd\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%xmm\[0-9\]+\[^\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)"  1 } } */
+/* { dg-final { scan-assembler-times "vreducesd\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%xmm\[0-9\]+\[^\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vreducesd\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%xmm\[0-9\]+\[^\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}\{z\}(?:\n|\[ \\t\]+#)" 1 } } */
+
+
 #include <immintrin.h>
 
+#define IMM 123
+
 volatile __m128d x1, x2;
 volatile __mmask8 m;
 
 void extern
 avx512dq_test (void)
 {
-  x1 = _mm_reduce_sd (x1, x2, 123);
+  x1 = _mm_reduce_sd (x1, x2, IMM);
+
+  x1 = _mm_mask_reduce_sd(x1, m, x1, x2, IMM);
+
+  x1 = _mm_maskz_reduce_sd(m, x1, x2, IMM);
 }

gcc/testsuite/gcc.target/i386/avx512dq-vreducesd-2.c

+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512dq" } */
+/* { dg-require-effective-target avx512dq } */
+
+#define AVX512DQ
+#include "avx512f-helper.h"
+#include <string.h>
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+#define IMM 0x23
+
+void
+CALC (double *r, double *s)
+{
+  int i;
+
+  memcpy (&r[1], &s[1], sizeof(double));
+
+  for (i = 0; i < 1; i++)
+    {
+      double tmp = (int) (4 * s[i]) / 4.0;
+      r[i] = s[i] - tmp;
+    }
+}
+
+void
+TEST (void)
+{
+  union128d res1, res2, res3;
+  union128d s1, s2, src;
+  double res_ref[2];
+  MASK_TYPE mask = MASK_VALUE;
+  int j;
+
+  for (j = 0; j < 2; j++)
+    {
+      s1.a[j] = j / 123.456;
+      s2.a[j] = j / 123.456;
+      res_ref[j] = j / 123.456;
+      res1.a[j] = DEFAULT_VALUE;
+      res2.a[j] = DEFAULT_VALUE;
+      res3.a[j] = DEFAULT_VALUE;
+    }
+
+  res1.x = _mm_reduce_sd (s1.x, s2.x, IMM);
+  res2.x = _mm_mask_reduce_sd (s1.x, mask, s1.x, s2.x, IMM);
+  res3.x = _mm_maskz_reduce_sd (mask, s1.x, s2.x, IMM);
+
+  CALC (res_ref, s2.a);
+
+  if (check_union128d (res1, res_ref))
+    abort ();
+ 
+  MASK_MERGE (d) (res_ref, mask, 1);
+
+  if (check_union128d (res2, res_ref))
+    abort ();
+
+  MASK_ZERO (d) (res_ref, mask, 1);
+
+  if (check_union128d (res3, res_ref))
+    abort ();
+
+}

gcc/testsuite/gcc.target/i386/avx512dq-vreducess-1.c

@@ -2,13 +2,23 @@
 /* { dg-options "-mavx512dq -O2" } */
 /* { dg-final { scan-assembler-times "vreducess\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
 
+/* { dg-final { scan-assembler-times "vreducess\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%xmm\[0-9\]+\[^\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)"  1 } } */
+/* { dg-final { scan-assembler-times "vreducess\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%xmm\[0-9\]+\[^\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vreducess\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%xmm\[0-9\]+\[^\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}\{z\}(?:\n|\[ \\t\]+#)" 1 } } */
+
 #include <immintrin.h>
 
+#define IMM 123
+
 volatile __m128 x1, x2;
 volatile __mmask8 m;
 
 void extern
 avx512dq_test (void)
 {
-  x1 = _mm_reduce_ss (x1, x2, 123);
+  x1 = _mm_reduce_ss (x1, x2, IMM);
+
+  x1 = _mm_mask_reduce_ss (x1, m, x1, x2, IMM);
+ 
+  x1 = _mm_maskz_reduce_ss (m, x1, x2, IMM);
 }

gcc/testsuite/gcc.target/i386/avx512dq-vreducess-2.c

+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512dq" } */
+/* { dg-require-effective-target avx512dq } */
+
+#define AVX512DQ
+#include "avx512f-helper.h"
+#include <string.h>
+
+#define SIZE (AVX512F_LEN / 64)
+#include "avx512f-mask-type.h"
+
+#define IMM 0x23
+
+void
+CALC (float *r, float *s)
+{
+  int i;
+
+  memcpy (&r[1], &s[1], 2 * sizeof(float));
+
+  for (i = 0; i < 2; i++)
+    {
+      float tmp = (int) (4 * s[i]) / 4.0;
+      r[i] = s[i] - tmp;
+    }
+}
+
+void
+TEST (void)
+{
+  printf("\nsize = %d\n\n", SIZE);
+
+  union128 res1, res2, res3;
+  union128 s1, s2, src;
+  float res_ref[4];
+  MASK_TYPE mask = MASK_VALUE;
+  int j;
+
+  for (j = 0; j < 4; j++)
+    {
+      s1.a[j] = j / 123.456;
+      s2.a[j] = j / 123.456;
+      res_ref[j] = j / 123.456;
+      res1.a[j] = DEFAULT_VALUE;
+      res2.a[j] = DEFAULT_VALUE;
+      res3.a[j] = DEFAULT_VALUE;
+    }
+
+  res1.x = _mm_reduce_ss (s1.x, s2.x, IMM);
+  res2.x = _mm_mask_reduce_ss (s1.x, mask, s1.x, s2.x, IMM);
+  res3.x = _mm_maskz_reduce_ss (mask, s1.x, s2.x, IMM);
+
+  CALC (res_ref, s2.a);
+
+  if (check_union128 (res1, res_ref))
+    abort ();
+ 
+  MASK_MERGE () (res_ref, mask, 1);
+
+  if (check_union128 (res2, res_ref))
+    abort ();
+
+  MASK_ZERO () (res_ref, mask, 1);
+
+  if (check_union128 (res3, res_ref))
+    abort ();
+
+}

gcc/testsuite/gcc.target/i386/sse-13.c

@@ -429,8 +429,8 @@
 /* avx512dqintrin.h */
 #define __builtin_ia32_kshiftliqi(A, B) __builtin_ia32_kshiftliqi(A, 8)
 #define __builtin_ia32_kshiftriqi(A, B) __builtin_ia32_kshiftriqi(A, 8)
-#define __builtin_ia32_reducess(A, B, F) __builtin_ia32_reducess(A, B, 1)
-#define __builtin_ia32_reducesd(A, B, F) __builtin_ia32_reducesd(A, B, 1)
+#define __builtin_ia32_reducess_mask(A, B, F, W, U) __builtin_ia32_reducess_mask(A, B, 1, W, U)
+#define __builtin_ia32_reducesd_mask(A, B, F, W, U) __builtin_ia32_reducesd_mask(A, B, 1, W, U)
 #define __builtin_ia32_reduceps512_mask(A, E, C, D) __builtin_ia32_reduceps512_mask(A, 1, C, D)
 #define __builtin_ia32_reducepd512_mask(A, E, C, D) __builtin_ia32_reducepd512_mask(A, 1, C, D)
 #define __builtin_ia32_rangess128_round(A, B, I, F) __builtin_ia32_rangess128_round(A, B, 1, 8)

gcc/testsuite/gcc.target/i386/sse-23.c

@@ -428,8 +428,8 @@
 /* avx512dqintrin.h */
 #define __builtin_ia32_kshiftliqi(A, B) __builtin_ia32_kshiftliqi(A, 8)
 #define __builtin_ia32_kshiftriqi(A, B) __builtin_ia32_kshiftriqi(A, 8)
-#define __builtin_ia32_reducess(A, B, F) __builtin_ia32_reducess(A, B, 1)
-#define __builtin_ia32_reducesd(A, B, F) __builtin_ia32_reducesd(A, B, 1)
+#define __builtin_ia32_reducess_mask(A, B, F, W, U) __builtin_ia32_reducess_mask(A, B, 1, W, U)
+#define __builtin_ia32_reducesd_mask(A, B, F, W, U) __builtin_ia32_reducesd_mask(A, B, 1, W, U)
 #define __builtin_ia32_reduceps512_mask(A, E, C, D) __builtin_ia32_reduceps512_mask(A, 1, C, D)
 #define __builtin_ia32_reducepd512_mask(A, E, C, D) __builtin_ia32_reducepd512_mask(A, 1, C, D)
 #define __builtin_ia32_rangess128_round(A, B, I, F) __builtin_ia32_rangess128_round(A, B, 1, 8)