https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88361
--- Comment #1 from Daniel Fruzynski <bugzi...@poradnik-webmastera.com> ---
For reference, this is NEON code which I used on AARCH64:
[code]
void test2()
{
int n = 0;
for (; n < SIZE*SIZE-3; n += 4)
{
// Copy data
uint32x4_t v = vld1q_u32((uint32_t*)(&src[0][0] + n));
vst1q_u32((uint32_t*)(&dst1[0][0] + n), v);
// Calculate bitmasks
v = vshlq_u32(vdupq_n_u32(1), vreinterpretq_s32_u32(v));
vst1q_u32((uint32_t*)(&dst2[0][0] + n), v);
}
for (; n < SIZE*SIZE; n++)
{
int x = *(&src[0][0] + n);
*((&dst1[0][0] + n)) = x;
*((&dst2[0][0] + n)) = 1 << x;
}
}
[/code]
