| 1 |
/* { dg-do compile } */ |
| 2 |
/* { dg-options "-mrvv-vector-bits=scalable -march=rv32gcv -mabi=ilp32 -fno-schedule-insns -fno-schedule-insns2" } */ |
| 3 |
|
| 4 |
#include "riscv_vector.h" |
| 5 |
|
| 6 |
void f(void *base, void *out, void *mask_in, size_t m, size_t n) { |
| 7 |
|
| 8 |
size_t vl = 107; |
| 9 |
for (size_t i = 0; i < m; i++) { |
| 10 |
if (i % 2 == 0) { |
| 11 |
for (size_t j = 0; j < n; j++){ |
| 12 |
vint8mf8_t v0 = __riscv_vle8_v_i8mf8(base + i + j + 700, vl); |
| 13 |
vint8mf8_t v1 = __riscv_vle8_v_i8mf8_tu(v0, base + i + j + 700, vl); |
| 14 |
__riscv_vse8_v_i8mf8 (out + i + j + 700, v1, vl); |
| 15 |
if (j % 2 == 0) { |
| 16 |
vint8mf8_t v0 = __riscv_vle8_v_i8mf8(base + i + j + 500, vl); |
| 17 |
__riscv_vse8_v_i8mf8 (out + i + j + 500, v0, vl); |
| 18 |
} else { |
| 19 |
vint16mf4_t v0 = __riscv_vle16_v_i16mf4(base + i + j + 600, vl); |
| 20 |
__riscv_vse16_v_i16mf4 (out + i + j + 600, v0, vl); |
| 21 |
} |
| 22 |
} |
| 23 |
} else { |
| 24 |
for (size_t j = 0; j < n; j++){ |
| 25 |
vint8mf8_t v0 = __riscv_vle8_v_i8mf8(base + i + j + 200, vl); |
| 26 |
vint8mf8_t v1 = __riscv_vle8_v_i8mf8_tu(v0, base + i + j + 300, vl); |
| 27 |
__riscv_vse8_v_i8mf8 (out + i + j + 400, v1, vl); |
| 28 |
} |
| 29 |
} |
| 30 |
} |
| 31 |
} |
| 32 |
|
| 33 |
/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*[a-x0-9]+,\s*e8,\s*mf8,\s*tu,\s*m[au]} 1 { target { no-opts "-O0" no-opts "-g" no-opts "-funroll-loops" } } } } */ |
| 34 |
/* { dg-final { scan-assembler-times {vsetvli} 1 { target { no-opts "-O0" no-opts "-g" no-opts "-funroll-loops" } } } } */ |