| 1 |
/* { dg-do compile } */ |
| 2 |
/* { dg-options "-mrvv-vector-bits=scalable -march=rv64gcv -mabi=lp64d -O3" } */ |
| 3 |
|
| 4 |
#include "riscv_vector.h" |
| 5 |
|
| 6 |
void |
| 7 |
foo (float *r, const float *x) |
| 8 |
{ |
| 9 |
int i, k; |
| 10 |
|
| 11 |
vfloat32m4_t x_vec; |
| 12 |
vfloat32m4_t x_forward_vec; |
| 13 |
vfloat32m4_t temp_vec; |
| 14 |
vfloat32m1_t dst_vec; |
| 15 |
vfloat32m1_t src_vec; |
| 16 |
|
| 17 |
float result = 0.0f; |
| 18 |
float shift_prev = 0.0f; |
| 19 |
|
| 20 |
size_t n = 64; |
| 21 |
for (size_t vl; n > 0; n -= vl) |
| 22 |
{ |
| 23 |
vl = __riscv_vsetvl_e32m4 (n); |
| 24 |
x_vec = __riscv_vle32_v_f32m4 (&x[0], vl); |
| 25 |
x_forward_vec = __riscv_vle32_v_f32m4 (&x[0], vl); |
| 26 |
temp_vec = __riscv_vfmul_vv_f32m4 (x_vec, x_forward_vec, vl); |
| 27 |
src_vec = __riscv_vfmv_s_tu (src_vec, 0.0f, vl); |
| 28 |
dst_vec = __riscv_vfmv_s_tu (dst_vec, 0.0f, vl); |
| 29 |
dst_vec = __riscv_vfredosum_tu (dst_vec, temp_vec, src_vec, vl); |
| 30 |
r[0] = __riscv_vfmv_f_s_f32m1_f32 (dst_vec); |
| 31 |
} |
| 32 |
} |
| 33 |
|
| 34 |
/* { dg-final { scan-assembler-times {vsetvli} 1 } } */ |
| 35 |
/* { dg-final { scan-assembler-not {vsetivli} } } */ |
| 36 |
/* { dg-final { scan-assembler-times {vsetvli\t[a-x0-9]+,\s*[a-x0-9]+,\s*e32,\s*m4,\s*tu,\s*m[au]} 1 } } */ |