* [PATCH][x86] Add missing mask intrinsics for MAX[SD,SS] and MIN[SD,SS]
@ 2017-05-30 14:30 Peryt, Sebastian
2017-05-31 18:37 ` Uros Bizjak
0 siblings, 1 reply; 2+ messages in thread
From: Peryt, Sebastian @ 2017-05-30 14:30 UTC (permalink / raw)
To: gcc-patches; +Cc: Uros Bizjak
[-- Attachment #1: Type: text/plain, Size: 1313 bytes --]
Hi,
This patch adds missing intrinsics for MAX[SD,SS] and MIN[SD,SS] listed below:
- _mm_mask_max_sd,
- _mm_maskz_max_sd,
- _mm_mask_max_ss,
- _mm_maskz_max_ss,
- _mm_mask_min_sd,
- _mm_maskz_min_sd,
- _mm_mask_min_ss,
- _mm_maskz_min_ss.
gcc/
* config/i386/avx512fintrin.h (_mm_mask_max_sd,
_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss,
_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss,
_mm_maskz_min_ss): New intrinsics.
gcc/testsuite/
* gcc.target/i386/avx512f-vmaxsd-1.c (_mm_mask_max_sd,
_mm_maskz_max_sd): Test new intrinsics.
* gcc.target/i386/avx512f-vmaxsd-2.c (_mm_mask_max_sd,
_mm_maskz_max_sd): Test new intrinsics.
* gcc.target/i386/avx512f-vmaxss-1.c (_mm_mask_max_ss,
_mm_maskz_max_ss): Test new intrinsics.
* gcc.target/i386/avx512f-vmaxss-2.c (_mm_mask_max_ss,
_mm_maskz_max_ss): Test new intrinsics.
* gcc.target/i386/avx512f-vminsd-1.c (_mm_mask_min_sd,
_mm_maskz_min_sd): Test new intrinsics.
* gcc.target/i386/avx512f-vminsd-2.c (_mm_mask_min_sd,
_mm_maskz_min_sd): Test new intrinsics.
* gcc.target/i386/avx512f-vminss-1.c (_mm_mask_min_ss,
_mm_maskz_min_ss): Test new intrinsics.
* gcc.target/i386/avx512f-vminss-2.c (_mm_mask_min_ss,
_mm_maskz_min_ss): Test new intrinsics.
Is it ok for trunk?
Thanks,
Sebastian
[-- Attachment #2: MASK_MAX[SD,SS]_MIN[SD,SS].patch --]
[-- Type: application/octet-stream, Size: 16419 bytes --]
diff --git a/gcc/config/i386/avx512fintrin.h b/gcc/config/i386/avx512fintrin.h
index acbf34f..287c8b8 100644
--- a/gcc/config/i386/avx512fintrin.h
+++ b/gcc/config/i386/avx512fintrin.h
@@ -11823,6 +11823,52 @@ _mm512_maskz_max_ps (__mmask16 __U, __m512 __A, __m512 __B)
_MM_FROUND_CUR_DIRECTION);
}
+extern __inline __m128d
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_max_sd (__m128d __W, __mmask8 __U, __m128d __A, __m128d __B)
+{
+ return (__m128d) __builtin_ia32_maxsd_mask_round ((__v2df) __A,
+ (__v2df) __B,
+ (__v2df) __W,
+ (__mmask8) __U,
+ _MM_FROUND_CUR_DIRECTION);
+}
+
+extern __inline __m128d
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_maskz_max_sd (__mmask8 __U, __m128d __A, __m128d __B)
+{
+ return (__m128d) __builtin_ia32_maxsd_mask_round ((__v2df) __A,
+ (__v2df) __B,
+ (__v2df)
+ _mm_setzero_pd (),
+ (__mmask8) __U,
+ _MM_FROUND_CUR_DIRECTION);
+}
+
+extern __inline __m128
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_max_ss (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B)
+{
+ return (__m128) __builtin_ia32_maxss_mask_round ((__v4sf) __A,
+ (__v4sf) __B,
+ (__v4sf) __W,
+ (__mmask8) __U,
+ _MM_FROUND_CUR_DIRECTION);
+}
+
+extern __inline __m128
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_maskz_max_ss (__mmask8 __U, __m128 __A, __m128 __B)
+{
+ return (__m128) __builtin_ia32_maxss_mask_round ((__v4sf) __A,
+ (__v4sf) __B,
+ (__v4sf)
+ _mm_setzero_ps (),
+ (__mmask8) __U,
+ _MM_FROUND_CUR_DIRECTION);
+}
+
extern __inline __m512d
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm512_min_pd (__m512d __A, __m512d __B)
@@ -11893,6 +11939,52 @@ _mm512_maskz_min_ps (__mmask16 __U, __m512 __A, __m512 __B)
_MM_FROUND_CUR_DIRECTION);
}
+extern __inline __m128d
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_min_sd (__m128d __W, __mmask8 __U, __m128d __A, __m128d __B)
+{
+ return (__m128d) __builtin_ia32_minsd_mask_round ((__v2df) __A,
+ (__v2df) __B,
+ (__v2df) __W,
+ (__mmask8) __U,
+ _MM_FROUND_CUR_DIRECTION);
+}
+
+extern __inline __m128d
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_maskz_min_sd (__mmask8 __U, __m128d __A, __m128d __B)
+{
+ return (__m128d) __builtin_ia32_minsd_mask_round ((__v2df) __A,
+ (__v2df) __B,
+ (__v2df)
+ _mm_setzero_pd (),
+ (__mmask8) __U,
+ _MM_FROUND_CUR_DIRECTION);
+}
+
+extern __inline __m128
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_min_ss (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B)
+{
+ return (__m128) __builtin_ia32_minss_mask_round ((__v4sf) __A,
+ (__v4sf) __B,
+ (__v4sf) __W,
+ (__mmask8) __U,
+ _MM_FROUND_CUR_DIRECTION);
+}
+
+extern __inline __m128
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_maskz_min_ss (__mmask8 __U, __m128 __A, __m128 __B)
+{
+ return (__m128) __builtin_ia32_minss_mask_round ((__v4sf) __A,
+ (__v4sf) __B,
+ (__v4sf)
+ _mm_setzero_ps (),
+ (__mmask8) __U,
+ _MM_FROUND_CUR_DIRECTION);
+}
+
extern __inline __m512d
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm512_scalef_pd (__m512d __A, __m512d __B)
diff --git a/gcc/testsuite/gcc.target/i386/avx512f-vmaxsd-1.c b/gcc/testsuite/gcc.target/i386/avx512f-vmaxsd-1.c
index c1c8f8d..161de34 100644
--- a/gcc/testsuite/gcc.target/i386/avx512f-vmaxsd-1.c
+++ b/gcc/testsuite/gcc.target/i386/avx512f-vmaxsd-1.c
@@ -1,5 +1,7 @@
/* { dg-do compile } */
/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vmaxsd\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vmaxsd\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}\{z\}(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vmaxsd\[ \\t\]+\[^\{\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vmaxsd\[ \\t\]+\[^\n\]*\{sae\}\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vmaxsd\[ \\t\]+\[^\n\]*\{sae\}\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}\{z\}(?:\n|\[ \\t\]+#)" 1 } } */
@@ -12,6 +14,8 @@ volatile __mmask8 m;
void extern
avx512f_test (void)
{
+ x1 = _mm_mask_max_sd (x1, m, x2, x3);
+ x1 = _mm_maskz_max_sd (m, x1, x2);
x1 = _mm_max_round_sd (x1, x2, _MM_FROUND_NO_EXC);
x1 = _mm_mask_max_round_sd (x1, m, x2, x3, _MM_FROUND_NO_EXC);
x1 = _mm_maskz_max_round_sd (m, x1, x2, _MM_FROUND_NO_EXC);
diff --git a/gcc/testsuite/gcc.target/i386/avx512f-vmaxsd-2.c b/gcc/testsuite/gcc.target/i386/avx512f-vmaxsd-2.c
index 29db777..e732d44 100644
--- a/gcc/testsuite/gcc.target/i386/avx512f-vmaxsd-2.c
+++ b/gcc/testsuite/gcc.target/i386/avx512f-vmaxsd-2.c
@@ -20,7 +20,7 @@ void
avx512f_test (void)
{
int i, sign;
- union128d res1, res2, res3, src1, src2;
+ union128d res1, res2, res3, res4, res5, src1, src2;
MASK_TYPE mask = MASK_VALUE;
double res_ref[SIZE];
@@ -29,26 +29,38 @@ avx512f_test (void)
{
src1.a[i] = 1.5 + 34.67 * i * sign;
src2.a[i] = -22.17 * i * sign + 1.0;
+ res1.a[i] = DEFAULT_VALUE;
+ res4.a[i] = DEFAULT_VALUE;
sign = sign * -1;
}
- for (i = 0; i < SIZE; i++)
- res2.a[i] = DEFAULT_VALUE;
- res1.x = _mm_max_round_sd (src1.x, src2.x, _MM_FROUND_NO_EXC);
- res2.x = _mm_mask_max_round_sd (res2.x, mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
- res3.x = _mm_maskz_max_round_sd (mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
+ res1.x = _mm_mask_max_sd (res1.x, mask, src1.x, src2.x);
+ res2.x = _mm_maskz_max_sd (mask, src1.x, src2.x);
+ res3.x = _mm_max_round_sd (src1.x, src2.x, _MM_FROUND_NO_EXC);
+ res4.x = _mm_mask_max_round_sd (res4.x, mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
+ res5.x = _mm_maskz_max_round_sd (mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
calc_max (res_ref, src1.a, src2.a);
+ MASK_MERGE (d) (res_ref, mask, 1);
if (check_union128d (res1, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, 1);
+ if (check_union128d (res2, res_ref))
+ abort ();
+
+ calc_max (res_ref, src1.a, src2.a);
+
+ if (check_union128d (res3, res_ref))
abort();
MASK_MERGE (d) (res_ref, mask, 1);
- if (check_union128d (res2, res_ref))
+ if (check_union128d (res4, res_ref))
abort ();
MASK_ZERO (d) (res_ref, mask, 1);
- if (check_union128d (res3, res_ref))
+ if (check_union128d (res5, res_ref))
abort ();
}
diff --git a/gcc/testsuite/gcc.target/i386/avx512f-vmaxss-1.c b/gcc/testsuite/gcc.target/i386/avx512f-vmaxss-1.c
index ecd8757..668afad 100644
--- a/gcc/testsuite/gcc.target/i386/avx512f-vmaxss-1.c
+++ b/gcc/testsuite/gcc.target/i386/avx512f-vmaxss-1.c
@@ -1,5 +1,7 @@
/* { dg-do compile } */
/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vmaxss\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vmaxss\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}\{z\}(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vmaxss\[ \\t\]+\[^\{\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vmaxss\[ \\t\]+\[^\n\]*\{sae\}\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vmaxss\[ \\t\]+\[^\n\]*\{sae\}\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}\{z\}(?:\n|\[ \\t\]+#)" 1 } } */
@@ -12,6 +14,8 @@ volatile __mmask8 m;
void extern
avx512f_test (void)
{
+ x1 = _mm_mask_max_ss (x1, m, x2, x3);
+ x1 = _mm_maskz_max_ss (m, x1, x2);
x1 = _mm_max_round_ss (x1, x2, _MM_FROUND_NO_EXC);
x1 = _mm_mask_max_round_ss (x1, m, x2, x3, _MM_FROUND_NO_EXC);
x1 = _mm_maskz_max_round_ss (m, x1, x2, _MM_FROUND_NO_EXC);
diff --git a/gcc/testsuite/gcc.target/i386/avx512f-vmaxss-2.c b/gcc/testsuite/gcc.target/i386/avx512f-vmaxss-2.c
index 1932a6e..c63a459 100644
--- a/gcc/testsuite/gcc.target/i386/avx512f-vmaxss-2.c
+++ b/gcc/testsuite/gcc.target/i386/avx512f-vmaxss-2.c
@@ -24,7 +24,7 @@ void
avx512f_test (void)
{
int i, sign;
- union128 res1, res2, res3, src1, src2;
+ union128 res1, res2, res3, res4, res5, src1, src2;
MASK_TYPE mask = MASK_VALUE;
float res_ref[SIZE];
@@ -33,26 +33,38 @@ avx512f_test (void)
{
src1.a[i] = 1.5 + 34.67 * i * sign;
src2.a[i] = -22.17 * i * sign + 1.0;
+ res1.a[i] = DEFAULT_VALUE;
+ res4.a[i] = DEFAULT_VALUE;
sign = sign * -1;
}
- for (i = 0; i < SIZE; i++)
- res2.a[i] = DEFAULT_VALUE;
- res1.x = _mm_max_round_ss (src1.x, src2.x, _MM_FROUND_NO_EXC);
- res2.x = _mm_mask_max_round_ss (res2.x, mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
- res3.x = _mm_maskz_max_round_ss (mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
+ res1.x = _mm_mask_max_ss (res1.x, mask, src1.x, src2.x);
+ res2.x = _mm_maskz_max_ss (mask, src1.x, src2.x);
+ res3.x = _mm_max_round_ss (src1.x, src2.x, _MM_FROUND_NO_EXC);
+ res4.x = _mm_mask_max_round_ss (res4.x, mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
+ res5.x = _mm_maskz_max_round_ss (mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
calc_max (res_ref, src1.a, src2.a);
+ MASK_MERGE () (res_ref, mask, 1);
if (check_union128 (res1, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, 1);
+ if (check_union128 (res2, res_ref))
+ abort ();
+
+ calc_max (res_ref, src1.a, src2.a);
+
+ if (check_union128 (res3, res_ref))
abort();
MASK_MERGE () (res_ref, mask, 1);
- if (check_union128 (res2, res_ref))
+ if (check_union128 (res4, res_ref))
abort ();
MASK_ZERO () (res_ref, mask, 1);
- if (check_union128 (res3, res_ref))
+ if (check_union128 (res5, res_ref))
abort ();
}
diff --git a/gcc/testsuite/gcc.target/i386/avx512f-vminsd-1.c b/gcc/testsuite/gcc.target/i386/avx512f-vminsd-1.c
index 22ada87..58b213d 100644
--- a/gcc/testsuite/gcc.target/i386/avx512f-vminsd-1.c
+++ b/gcc/testsuite/gcc.target/i386/avx512f-vminsd-1.c
@@ -1,5 +1,7 @@
/* { dg-do compile } */
/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vminsd\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vminsd\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}\{z\}(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vminsd\[ \\t\]+\[^\{\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vminsd\[ \\t\]+\[^\n\]*\{sae\}\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vminsd\[ \\t\]+\[^\n\]*\{sae\}\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}\{z\}(?:\n|\[ \\t\]+#)" 1 } } */
@@ -12,6 +14,8 @@ volatile __mmask8 m;
void extern
avx512f_test (void)
{
+ x1 = _mm_mask_min_sd (x1, m, x2, x3);
+ x1 = _mm_maskz_min_sd (m, x1, x2);
x1 = _mm_min_round_sd (x1, x2, _MM_FROUND_NO_EXC);
x1 = _mm_mask_min_round_sd (x1, m, x2, x3, _MM_FROUND_NO_EXC);
x1 = _mm_maskz_min_round_sd (m, x1, x2, _MM_FROUND_NO_EXC);
diff --git a/gcc/testsuite/gcc.target/i386/avx512f-vminsd-2.c b/gcc/testsuite/gcc.target/i386/avx512f-vminsd-2.c
index 79a051f..464ce96 100644
--- a/gcc/testsuite/gcc.target/i386/avx512f-vminsd-2.c
+++ b/gcc/testsuite/gcc.target/i386/avx512f-vminsd-2.c
@@ -20,7 +20,7 @@ void
avx512f_test (void)
{
int i, sign;
- union128d res1, res2, res3, src1, src2;
+ union128d res1, res2, res3, res4, res5, src1, src2;
MASK_TYPE mask = MASK_VALUE;
double res_ref[SIZE];
@@ -29,26 +29,38 @@ avx512f_test (void)
{
src1.a[i] = 1.5 + 34.67 * i * sign;
src2.a[i] = -22.17 * i * sign + 1.0;
+ res1.a[i] = DEFAULT_VALUE;
+ res4.a[i] = DEFAULT_VALUE;
sign = sign * -1;
}
- for (i = 0; i < SIZE; i++)
- res2.a[i] = DEFAULT_VALUE;
-
- res1.x = _mm_min_round_sd (src1.x, src2.x, _MM_FROUND_NO_EXC);
- res2.x = _mm_mask_min_round_sd (res2.x, mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
- res3.x = _mm_maskz_min_round_sd (mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
+
+ res1.x = _mm_mask_min_sd (res1.x, mask, src1.x, src2.x);
+ res2.x = _mm_maskz_min_sd (mask, src1.x, src2.x);
+ res3.x = _mm_min_round_sd (src1.x, src2.x, _MM_FROUND_NO_EXC);
+ res4.x = _mm_mask_min_round_sd (res4.x, mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
+ res5.x = _mm_maskz_min_round_sd (mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
calc_min (res_ref, src1.a, src2.a);
+ MASK_MERGE (d) (res_ref, mask, 1);
if (check_union128d (res1, res_ref))
+ abort ();
+
+ MASK_ZERO (d) (res_ref, mask, 1);
+ if (check_union128d (res2, res_ref))
+ abort ();
+
+ calc_min (res_ref, src1.a, src2.a);
+
+ if (check_union128d (res3, res_ref))
abort();
MASK_MERGE (d) (res_ref, mask, 1);
- if (check_union128d (res2, res_ref))
+ if (check_union128d (res4, res_ref))
abort ();
MASK_ZERO (d) (res_ref, mask, 1);
- if (check_union128d (res3, res_ref))
+ if (check_union128d (res5, res_ref))
abort ();
}
diff --git a/gcc/testsuite/gcc.target/i386/avx512f-vminss-1.c b/gcc/testsuite/gcc.target/i386/avx512f-vminss-1.c
index 5703a88..7d2175e 100644
--- a/gcc/testsuite/gcc.target/i386/avx512f-vminss-1.c
+++ b/gcc/testsuite/gcc.target/i386/avx512f-vminss-1.c
@@ -1,5 +1,7 @@
/* { dg-do compile } */
/* { dg-options "-mavx512f -O2" } */
+/* { dg-final { scan-assembler-times "vminss\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vminss\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}\{z\}(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vminss\[ \\t\]+\[^\{\n\]*\{sae\}\[^\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vminss\[ \\t\]+\[^\n\]*\{sae\}\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vminss\[ \\t\]+\[^\n\]*\{sae\}\[^\{\n\]*%xmm\[0-9\]+\{%k\[1-7\]\}\{z\}(?:\n|\[ \\t\]+#)" 1 } } */
@@ -12,6 +14,8 @@ volatile __mmask8 m;
void extern
avx512f_test (void)
{
+ x1 = _mm_mask_min_ss (x1, m, x2, x3);
+ x1 = _mm_maskz_min_ss (m, x1, x2);
x1 = _mm_min_round_ss (x1, x2, _MM_FROUND_NO_EXC);
x1 = _mm_mask_min_round_ss (x1, m, x2, x3, _MM_FROUND_NO_EXC);
x1 = _mm_maskz_min_round_ss (m, x1, x2, _MM_FROUND_NO_EXC);
diff --git a/gcc/testsuite/gcc.target/i386/avx512f-vminss-2.c b/gcc/testsuite/gcc.target/i386/avx512f-vminss-2.c
index ad1fcb9..351c273 100644
--- a/gcc/testsuite/gcc.target/i386/avx512f-vminss-2.c
+++ b/gcc/testsuite/gcc.target/i386/avx512f-vminss-2.c
@@ -24,7 +24,7 @@ void
avx512f_test (void)
{
int i, sign;
- union128 res1, res2, res3, src1, src2;
+ union128 res1, res2, res3, res4, res5, src1, src2;
MASK_TYPE mask = MASK_VALUE;
float res_ref[SIZE];
@@ -33,26 +33,38 @@ avx512f_test (void)
{
src1.a[i] = 1.5 + 34.67 * i * sign;
src2.a[i] = -22.17 * i * sign + 1.0;
+ res1.a[i] = DEFAULT_VALUE;
+ res4.a[i] = DEFAULT_VALUE;
sign = sign * -1;
}
- for (i = 0; i < SIZE; i++)
- res2.a[i] = DEFAULT_VALUE;
- res1.x = _mm_min_round_ss (src1.x, src2.x, _MM_FROUND_NO_EXC);
- res2.x = _mm_mask_min_round_ss (res2.x, mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
- res3.x = _mm_maskz_min_round_ss (mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
+ res1.x = _mm_mask_min_ss (res1.x, mask, src1.x, src2.x);
+ res2.x = _mm_maskz_min_ss (mask, src1.x, src2.x);
+ res3.x = _mm_min_round_ss (src1.x, src2.x, _MM_FROUND_NO_EXC);
+ res4.x = _mm_mask_min_round_ss (res4.x, mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
+ res5.x = _mm_maskz_min_round_ss (mask, src1.x, src2.x, _MM_FROUND_NO_EXC);
calc_min (res_ref, src1.a, src2.a);
-
+
+ MASK_MERGE () (res_ref, mask, 1);
if (check_union128 (res1, res_ref))
+ abort ();
+
+ MASK_ZERO () (res_ref, mask, 1);
+ if (check_union128 (res2, res_ref))
+ abort ();
+
+ calc_min (res_ref, src1.a, src2.a);
+
+ if (check_union128 (res3, res_ref))
abort();
MASK_MERGE () (res_ref, mask, 1);
- if (check_union128 (res2, res_ref))
+ if (check_union128 (res4, res_ref))
abort ();
MASK_ZERO () (res_ref, mask, 1);
- if (check_union128 (res3, res_ref))
+ if (check_union128 (res5, res_ref))
abort ();
}
^ permalink raw reply [flat|nested] 2+ messages in thread* Re: [PATCH][x86] Add missing mask intrinsics for MAX[SD,SS] and MIN[SD,SS]
2017-05-30 14:30 [PATCH][x86] Add missing mask intrinsics for MAX[SD,SS] and MIN[SD,SS] Peryt, Sebastian
@ 2017-05-31 18:37 ` Uros Bizjak
0 siblings, 0 replies; 2+ messages in thread
From: Uros Bizjak @ 2017-05-31 18:37 UTC (permalink / raw)
To: Peryt, Sebastian; +Cc: gcc-patches
On Tue, May 30, 2017 at 4:29 PM, Peryt, Sebastian
<sebastian.peryt@intel.com> wrote:
> Hi,
>
> This patch adds missing intrinsics for MAX[SD,SS] and MIN[SD,SS] listed below:
> - _mm_mask_max_sd,
> - _mm_maskz_max_sd,
> - _mm_mask_max_ss,
> - _mm_maskz_max_ss,
>
> - _mm_mask_min_sd,
> - _mm_maskz_min_sd,
> - _mm_mask_min_ss,
> - _mm_maskz_min_ss.
>
> gcc/
> * config/i386/avx512fintrin.h (_mm_mask_max_sd,
> _mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss,
> _mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss,
> _mm_maskz_min_ss): New intrinsics.
>
> gcc/testsuite/
> * gcc.target/i386/avx512f-vmaxsd-1.c (_mm_mask_max_sd,
> _mm_maskz_max_sd): Test new intrinsics.
> * gcc.target/i386/avx512f-vmaxsd-2.c (_mm_mask_max_sd,
> _mm_maskz_max_sd): Test new intrinsics.
> * gcc.target/i386/avx512f-vmaxss-1.c (_mm_mask_max_ss,
> _mm_maskz_max_ss): Test new intrinsics.
> * gcc.target/i386/avx512f-vmaxss-2.c (_mm_mask_max_ss,
> _mm_maskz_max_ss): Test new intrinsics.
> * gcc.target/i386/avx512f-vminsd-1.c (_mm_mask_min_sd,
> _mm_maskz_min_sd): Test new intrinsics.
> * gcc.target/i386/avx512f-vminsd-2.c (_mm_mask_min_sd,
> _mm_maskz_min_sd): Test new intrinsics.
> * gcc.target/i386/avx512f-vminss-1.c (_mm_mask_min_ss,
> _mm_maskz_min_ss): Test new intrinsics.
> * gcc.target/i386/avx512f-vminss-2.c (_mm_mask_min_ss,
> _mm_maskz_min_ss): Test new intrinsics.
>
> Is it ok for trunk?
Approved and committed to mainline SVN.
Thanks,
Uros.
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2017-05-31 18:27 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-05-30 14:30 [PATCH][x86] Add missing mask intrinsics for MAX[SD,SS] and MIN[SD,SS] Peryt, Sebastian
2017-05-31 18:37 ` Uros Bizjak
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).