ligang/vpp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Enable NEON instructions in memcpy_le

Browse Source 
Neon version of memcpy_le gives better performance compared with memmove on aarch64

Change-Id: I44b487bb0795a6e70dd1e55bdde4a077773ec859
Signed-off-by: Lijian Zhang <Lijian.Zhang@arm.com>
Reviewed-by: Sirshak Das <sirdas@arm.com>
This commit is contained in:
Lijian.Zhang
2019-04-04 15:26:26 +08:00
committed by Damjan Marion
parent 1cff643c65
commit 37c83785b6
2 changed files with 9 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/vppinfra/string.h
View File

@ -239,7 +239,7 @@ clib_memcpy_le (u8 * dst, u8 * src, u8 len, u8 max_len)
d1 = u8x32_blend (d1, s1, u8x32_is_greater (lv, mask));
u8x32_store_unaligned (d1, dst + 32);
#elif defined (CLIB_HAVE_VEC128) && !defined (__aarch64__)
#elif defined (CLIB_HAVE_VEC128)
u8x16 s0, s1, s2, s3, d0, d1, d2, d3;
u8x16 mask = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 };
u8x16 lv = u8x16_splat (len);

8
src/vppinfra/vector_neon.h
View File

@ -83,6 +83,14 @@ static_always_inline u32 \
t##s##x##c##_zero_byte_mask (t##s##x##c x) \
{ uint8x16_t v = vreinterpretq_u8_u##s (vceqq_##i (vdupq_n_##i(0), x)); \
return u8x16_compare_byte_mask (v); } \
\
static_always_inline u##s##x##c \
t##s##x##c##_is_greater (t##s##x##c a, t##s##x##c b) \
{ return (u##s##x##c) vcgtq_##i (a, b); } \
\
static_always_inline t##s##x##c \
t##s##x##c##_blend (t##s##x##c dst, t##s##x##c src, u##s##x##c mask) \
{ return (t##s##x##c) vbslq_##i (mask, src, dst); }
foreach_neon_vec128i foreach_neon_vec128u