1 /* Test that the compiler properly optimizes floating point multiply and add
2 instructions vector into fmaddps on SSE5 systems. */
4 /* { dg-do compile { target x86_64-*-*} } */
5 /* { dg-options "-O2 -msse5 -mfused-madd -ftree-vectorize" } */
7 extern void exit (int);
9 typedef float __m128 __attribute__ ((__vector_size__ (16), __may_alias__));
10 typedef double __m128d __attribute__ ((__vector_size__ (16), __may_alias__));
26 for (i = 0; i < SIZE; i++)
27 a.f[i] = (b.f[i] * c.f[i]) + d.f[i];
35 for (i = 0; i < SIZE; i++)
36 a.d[i] = (b.d[i] * c.d[i]) + d.d[i];
44 for (i = 0; i < SIZE; i++)
45 a.f[i] = (b.f[i] * c.f[i]) - d.f[i];
53 for (i = 0; i < SIZE; i++)
54 a.d[i] = (b.d[i] * c.d[i]) - d.d[i];
58 flt_neg_mul_add (void)
62 for (i = 0; i < SIZE; i++)
63 a.f[i] = (-(b.f[i] * c.f[i])) + d.f[i];
67 dbl_neg_mul_add (void)
71 for (i = 0; i < SIZE; i++)
72 a.d[i] = (-(b.d[i] * c.d[i])) + d.d[i];
87 /* { dg-final { scan-assembler "fmaddps" } } */
88 /* { dg-final { scan-assembler "fmaddpd" } } */
89 /* { dg-final { scan-assembler "fmsubps" } } */
90 /* { dg-final { scan-assembler "fmsubpd" } } */
91 /* { dg-final { scan-assembler "fnmaddps" } } */
92 /* { dg-final { scan-assembler "fnmaddpd" } } */