https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119187

--- Comment #15 from Tamar Christina <tnfchris at gcc dot gnu.org> ---
unrolling works too

ok, unrolling works too
```
.L5:
        ldr     q29, [x1, x4]
        ldr     q31, [x0, x4]
        ldr     q0, [x9, x4]
        ldr     q30, [x8, x4]
        cmhi    v31.16b, v31.16b, v29.16b
        cmhi    v30.16b, v30.16b, v0.16b
        str     q31, [x2, x4]
        str     q30, [x7, x4]
        add     x4, x4, 32
        cmp     x10, x4
        bne     .L5
```
from
```
#include <arm_neon.h>

void foo (uint8_t *a, uint8_t *b, uint8_t *c, int n)
{
#pragma GCC unroll 32
  for (int i = 0; i < n; i+=8, a+=8, b+=8, c+=8)
    {
        uint8x8_t av1 = vld1_u8 (a);
        uint8x8_t bv1 = vld1_u8 (b);
        vst1_u8 (c, vcgt_u8 (av1, bv1));
    }
}
```

so that's good.

BB SLP works too

#include <arm_neon.h>

void foo (uint8_t *a, uint8_t *b, uint8_t *c, int n)
{
  uint8x8_t av1 = vld1_u8 (a);
  uint8x8_t av2 = vld1_u8 (a+8);
  uint8x8_t bv1 = vld1_u8 (b);
  uint8x8_t bv2 = vld1_u8 (b+8);
  vst1_u8 (c, vadd_u8 (av1, bv1));
  vst1_u8 (c+8, vadd_u8 (av2, bv2));
}

turns into

foo:
        ldr     q31, [x0]
        ldr     q30, [x1]
        add     v30.16b, v31.16b, v30.16b
        str     q30, [x2]
        ret

it can't do SVE yet without Chris's BB VLA SLP support.

Reply via email to