| 1 |
318 |
jeremybenn |
/* Test whether using target specific options, we can generate FMA4 code. */
|
| 2 |
|
|
/* { dg-do compile } */
|
| 3 |
|
|
/* { dg-require-effective-target lp64 } */
|
| 4 |
|
|
/* { dg-options "-O2 -march=k8" } */
|
| 5 |
|
|
|
| 6 |
|
|
extern void exit (int);
|
| 7 |
|
|
|
| 8 |
|
|
#define FMA4_ATTR __attribute__((__target__("fma4")))
|
| 9 |
|
|
extern float flt_mul_add (float a, float b, float c) FMA4_ATTR;
|
| 10 |
|
|
extern float flt_mul_sub (float a, float b, float c) FMA4_ATTR;
|
| 11 |
|
|
extern float flt_neg_mul_add (float a, float b, float c) FMA4_ATTR;
|
| 12 |
|
|
extern float flt_neg_mul_sub (float a, float b, float c) FMA4_ATTR;
|
| 13 |
|
|
|
| 14 |
|
|
extern double dbl_mul_add (double a, double b, double c) FMA4_ATTR;
|
| 15 |
|
|
extern double dbl_mul_sub (double a, double b, double c) FMA4_ATTR;
|
| 16 |
|
|
extern double dbl_neg_mul_add (double a, double b, double c) FMA4_ATTR;
|
| 17 |
|
|
extern double dbl_neg_mul_sub (double a, double b, double c) FMA4_ATTR;
|
| 18 |
|
|
|
| 19 |
|
|
float
|
| 20 |
|
|
flt_mul_add (float a, float b, float c)
|
| 21 |
|
|
{
|
| 22 |
|
|
return (a * b) + c;
|
| 23 |
|
|
}
|
| 24 |
|
|
|
| 25 |
|
|
double
|
| 26 |
|
|
dbl_mul_add (double a, double b, double c)
|
| 27 |
|
|
{
|
| 28 |
|
|
return (a * b) + c;
|
| 29 |
|
|
}
|
| 30 |
|
|
|
| 31 |
|
|
float
|
| 32 |
|
|
flt_mul_sub (float a, float b, float c)
|
| 33 |
|
|
{
|
| 34 |
|
|
return (a * b) - c;
|
| 35 |
|
|
}
|
| 36 |
|
|
|
| 37 |
|
|
double
|
| 38 |
|
|
dbl_mul_sub (double a, double b, double c)
|
| 39 |
|
|
{
|
| 40 |
|
|
return (a * b) - c;
|
| 41 |
|
|
}
|
| 42 |
|
|
|
| 43 |
|
|
float
|
| 44 |
|
|
flt_neg_mul_add (float a, float b, float c)
|
| 45 |
|
|
{
|
| 46 |
|
|
return (-(a * b)) + c;
|
| 47 |
|
|
}
|
| 48 |
|
|
|
| 49 |
|
|
double
|
| 50 |
|
|
dbl_neg_mul_add (double a, double b, double c)
|
| 51 |
|
|
{
|
| 52 |
|
|
return (-(a * b)) + c;
|
| 53 |
|
|
}
|
| 54 |
|
|
|
| 55 |
|
|
float
|
| 56 |
|
|
flt_neg_mul_sub (float a, float b, float c)
|
| 57 |
|
|
{
|
| 58 |
|
|
return (-(a * b)) - c;
|
| 59 |
|
|
}
|
| 60 |
|
|
|
| 61 |
|
|
double
|
| 62 |
|
|
dbl_neg_mul_sub (double a, double b, double c)
|
| 63 |
|
|
{
|
| 64 |
|
|
return (-(a * b)) - c;
|
| 65 |
|
|
}
|
| 66 |
|
|
|
| 67 |
|
|
float f[10] = { 2, 3, 4 };
|
| 68 |
|
|
double d[10] = { 2, 3, 4 };
|
| 69 |
|
|
|
| 70 |
|
|
int main ()
|
| 71 |
|
|
{
|
| 72 |
|
|
f[3] = flt_mul_add (f[0], f[1], f[2]);
|
| 73 |
|
|
f[4] = flt_mul_sub (f[0], f[1], f[2]);
|
| 74 |
|
|
f[5] = flt_neg_mul_add (f[0], f[1], f[2]);
|
| 75 |
|
|
f[6] = flt_neg_mul_sub (f[0], f[1], f[2]);
|
| 76 |
|
|
|
| 77 |
|
|
d[3] = dbl_mul_add (d[0], d[1], d[2]);
|
| 78 |
|
|
d[4] = dbl_mul_sub (d[0], d[1], d[2]);
|
| 79 |
|
|
d[5] = dbl_neg_mul_add (d[0], d[1], d[2]);
|
| 80 |
|
|
d[6] = dbl_neg_mul_sub (d[0], d[1], d[2]);
|
| 81 |
|
|
exit (0);
|
| 82 |
|
|
}
|
| 83 |
|
|
|
| 84 |
|
|
/* { dg-final { scan-assembler "vfmaddss" } } */
|
| 85 |
|
|
/* { dg-final { scan-assembler "vfmaddsd" } } */
|
| 86 |
|
|
/* { dg-final { scan-assembler "vfmsubss" } } */
|
| 87 |
|
|
/* { dg-final { scan-assembler "vfmsubsd" } } */
|
| 88 |
|
|
/* { dg-final { scan-assembler "vfnmaddss" } } */
|
| 89 |
|
|
/* { dg-final { scan-assembler "vfnmaddsd" } } */
|
| 90 |
|
|
/* { dg-final { scan-assembler "vfnmsubss" } } */
|
| 91 |
|
|
/* { dg-final { scan-assembler "vfnmsubsd" } } */
|
| 92 |
|
|
/* { dg-final { scan-assembler "call\t(.*)flt_mul_add" } } */
|
| 93 |
|
|
/* { dg-final { scan-assembler "call\t(.*)flt_mul_sub" } } */
|
| 94 |
|
|
/* { dg-final { scan-assembler "call\t(.*)flt_neg_mul_add" } } */
|
| 95 |
|
|
/* { dg-final { scan-assembler "call\t(.*)flt_neg_mul_sub" } } */
|
| 96 |
|
|
/* { dg-final { scan-assembler "call\t(.*)dbl_mul_add" } } */
|
| 97 |
|
|
/* { dg-final { scan-assembler "call\t(.*)dbl_mul_sub" } } */
|
| 98 |
|
|
/* { dg-final { scan-assembler "call\t(.*)dbl_neg_mul_add" } } */
|
| 99 |
|
|
/* { dg-final { scan-assembler "call\t(.*)dbl_neg_mul_sub" } } */
|
