1 /* Test that the compiler properly optimizes floating point multiply and add
2 instructions vector into vfmaddps on FMA4 systems. */
4 /* { dg-do compile } */
5 /* { dg-require-effective-target lp64 } */
6 /* { dg-options "-O2 -mfma4 -ftree-vectorize" } */
8 extern void exit (int);
10 typedef float __m256 __attribute__ ((__vector_size__ (32), __may_alias__));
11 typedef double __m256d __attribute__ ((__vector_size__ (32), __may_alias__));
27 for (i = 0; i < SIZE; i++)
28 a.f[i] = (b.f[i] * c.f[i]) + d.f[i];
36 for (i = 0; i < SIZE; i++)
37 a.d[i] = (b.d[i] * c.d[i]) + d.d[i];
45 for (i = 0; i < SIZE; i++)
46 a.f[i] = (b.f[i] * c.f[i]) - d.f[i];
54 for (i = 0; i < SIZE; i++)
55 a.d[i] = (b.d[i] * c.d[i]) - d.d[i];
59 flt_neg_mul_add (void)
63 for (i = 0; i < SIZE; i++)
64 a.f[i] = (-(b.f[i] * c.f[i])) + d.f[i];
68 dbl_neg_mul_add (void)
72 for (i = 0; i < SIZE; i++)
73 a.d[i] = (-(b.d[i] * c.d[i])) + d.d[i];
88 /* { dg-final { scan-assembler "vfmaddps" } } */
89 /* { dg-final { scan-assembler "vfmaddpd" } } */
90 /* { dg-final { scan-assembler "vfmsubps" } } */
91 /* { dg-final { scan-assembler "vfmsubpd" } } */
92 /* { dg-final { scan-assembler "vfnmaddps" } } */
93 /* { dg-final { scan-assembler "vfnmaddpd" } } */