| 1 |
/* { dg-options "-O2" } */ |
| 2 |
|
| 3 |
#include <arm_neon.h> |
| 4 |
typedef struct { |
| 5 |
float re; |
| 6 |
float im; |
| 7 |
} cmplx_f32_t; |
| 8 |
|
| 9 |
void test2x2_f32(const cmplx_f32_t *p_src_a, |
| 10 |
const cmplx_f32_t *p_src_b, |
| 11 |
cmplx_f32_t *p_dst) { |
| 12 |
const float32_t *a_ptr = (const float32_t *)p_src_a; |
| 13 |
const float32_t *b_ptr = (const float32_t *)p_src_b; |
| 14 |
float32_t *out_ptr = (float32_t *)p_dst; |
| 15 |
|
| 16 |
float32x2x2_t a_col[2]; |
| 17 |
float32x2x2_t b[2]; |
| 18 |
float32x2x2_t result[2]; |
| 19 |
|
| 20 |
a_col[0] = vld2_f32(a_ptr); |
| 21 |
b[0] = vld2_f32(b_ptr); |
| 22 |
|
| 23 |
result[0].val[0] = vmul_lane_f32(a_col[0].val[0], b[0].val[0], 0); |
| 24 |
result[0].val[1] = vmul_lane_f32(a_col[0].val[1], b[0].val[0], 0); |
| 25 |
|
| 26 |
vst2_f32(out_ptr, result[0]); |
| 27 |
out_ptr = out_ptr + 4; |
| 28 |
} |
| 29 |
|
| 30 |
/* { dg-final { scan-assembler-not {\tdup\t} } } */ |
| 31 |
/* { dg-final { scan-assembler-times {(?n)\tfmul\t.*v[0-9]+\.s\[0\]\n} 2 } } */ |