* gcc.target/i386/pr70021.c: Add -mtune=skylake.
[official-gcc.git] / gcc / testsuite / gcc.target / i386 / fma4-msubXX.c
blobeed75580e8d32e198749a9f4b856afa6bb743437
1 /* { dg-do run } */
2 /* { dg-require-effective-target fma4 } */
3 /* { dg-options "-O0 -mfma4" } */
5 #include "fma4-check.h"
7 #include <x86intrin.h>
8 #include <string.h>
10 #define NUM 20
12 union
14 __m128 x[NUM];
15 float f[NUM * 4];
16 __m128d y[NUM];
17 double d[NUM * 2];
18 } dst, res, src1, src2, src3;
20 /* Note that in macc*,msub*,mnmacc* and mnsub* instructions, the intermdediate
21 product is not rounded, only the addition is rounded. */
23 static void
24 init_msubps ()
26 int i;
27 for (i = 0; i < NUM * 4; i++)
29 src1.f[i] = i;
30 src2.f[i] = i + 10;
31 src3.f[i] = i + 20;
35 static void
36 init_msubpd ()
38 int i;
39 for (i = 0; i < NUM * 4; i++)
41 src1.d[i] = i;
42 src2.d[i] = i + 10;
43 src3.d[i] = i + 20;
47 static int
48 check_msubps ()
50 int i, j, check_fails = 0;
51 for (i = 0; i < NUM * 4; i = i + 4)
52 for (j = 0; j < 4; j++)
54 res.f[i + j] = (src1.f[i + j] * src2.f[i + j]) - src3.f[i + j];
55 if (dst.f[i + j] != res.f[i + j])
56 check_fails++;
58 return check_fails++;
61 static int
62 check_msubpd ()
64 int i, j, check_fails = 0;
65 for (i = 0; i < NUM * 2; i = i + 2)
66 for (j = 0; j < 2; j++)
68 res.d[i + j] = (src1.d[i + j] * src2.d[i + j]) - src3.d[i + j];
69 if (dst.d[i + j] != res.d[i + j])
70 check_fails++;
72 return check_fails++;
76 static int
77 check_msubss ()
79 int i, j, check_fails = 0;
80 for (i = 0; i < NUM * 4; i = i + 4)
82 res.f[i] = (src1.f[i] * src2.f[i]) - src3.f[i];
83 if (dst.f[i] != res.f[i])
84 check_fails++;
86 return check_fails++;
89 static int
90 check_msubsd ()
92 int i, j, check_fails = 0;
93 for (i = 0; i < NUM * 2; i = i + 2)
95 res.d[i] = (src1.d[i] * src2.d[i]) - src3.d[i];
96 if (dst.d[i] != res.d[i])
97 check_fails++;
99 return check_fails++;
102 static void
103 fma4_test (void)
105 int i;
107 init_msubps ();
109 for (i = 0; i < NUM; i++)
110 dst.x[i] = _mm_msub_ps (src1.x[i], src2.x[i], src3.x[i]);
112 if (check_msubps ())
113 abort ();
115 for (i = 0; i < NUM; i++)
116 dst.x[i] = _mm_msub_ss (src1.x[i], src2.x[i], src3.x[i]);
118 if (check_msubss ())
119 abort ();
121 init_msubpd ();
123 for (i = 0; i < NUM; i++)
124 dst.y[i] = _mm_msub_pd (src1.y[i], src2.y[i], src3.y[i]);
126 if (check_msubpd ())
127 abort ();
129 for (i = 0; i < NUM; i++)
130 dst.y[i] = _mm_msub_sd (src1.y[i], src2.y[i], src3.y[i]);
132 if (check_msubsd ())
133 abort ();