| 1 |
/* { dg-do assemble { target aarch64_asm_sve_ok } } */ |
| 2 |
/* { dg-options "-O3 -msve-vector-bits=256 --save-temps" } */ |
| 3 |
|
| 4 |
typedef _Float16 vnx8hf __attribute__((vector_size(32))); |
| 5 |
typedef float vnx4sf __attribute__((vector_size(32))); |
| 6 |
typedef double vnx2df __attribute__((vector_size(32))); |
| 7 |
|
| 8 |
#define DO_OP(TYPE) \ |
| 9 |
void vdiv_##TYPE (TYPE *x, TYPE y) \ |
| 10 |
{ \ |
| 11 |
register TYPE dst asm("z0"); \ |
| 12 |
register TYPE src asm("z2"); \ |
| 13 |
dst = *x; \ |
| 14 |
src = y; \ |
| 15 |
asm volatile ("" :: "w" (dst), "w" (src)); \ |
| 16 |
dst = dst / src; \ |
| 17 |
asm volatile ("" :: "w" (dst)); \ |
| 18 |
*x = dst; \ |
| 19 |
} \ |
| 20 |
void vdivr_##TYPE (TYPE *x, TYPE y) \ |
| 21 |
{ \ |
| 22 |
register TYPE dst asm("z0"); \ |
| 23 |
register TYPE src asm("z2"); \ |
| 24 |
dst = *x; \ |
| 25 |
src = y; \ |
| 26 |
asm volatile ("" :: "w" (dst), "w" (src)); \ |
| 27 |
dst = src / dst; \ |
| 28 |
asm volatile ("" :: "w" (dst)); \ |
| 29 |
*x = dst; \ |
| 30 |
} |
| 31 |
|
| 32 |
DO_OP (vnx8hf) |
| 33 |
DO_OP (vnx4sf) |
| 34 |
DO_OP (vnx2df) |
| 35 |
|
| 36 |
/* { dg-final { scan-assembler-times {\tfdiv\tz[0-9]+\.h, p[0-7]/m, z[0-9]+\.h, z[0-9]+\.h\n} 1 } } */ |
| 37 |
/* { dg-final { scan-assembler-times {\tfdivr\tz[0-9]+\.h, p[0-7]/m, z[0-9]+\.h, z[0-9]+\.h\n} 1 } } */ |
| 38 |
|
| 39 |
/* { dg-final { scan-assembler-times {\tfdiv\tz[0-9]+\.s, p[0-7]/m, z[0-9]+\.s, z[0-9]+\.s\n} 1 } } */ |
| 40 |
/* { dg-final { scan-assembler-times {\tfdivr\tz[0-9]+\.s, p[0-7]/m, z[0-9]+\.s, z[0-9]+\.s\n} 1 } } */ |
| 41 |
|
| 42 |
/* { dg-final { scan-assembler-times {\tfdiv\tz[0-9]+\.d, p[0-7]/m, z[0-9]+\.d, z[0-9]+\.d\n} 1 } } */ |
| 43 |
/* { dg-final { scan-assembler-times {\tfdivr\tz[0-9]+\.d, p[0-7]/m, z[0-9]+\.d, z[0-9]+\.d\n} 1 } } */ |