hashirama/uvg266
1
0
Fork 0
mirror of https://github.com/ultravideo/uvg266.git synced 2024-11-27 19:24:06 +00:00
Code Issues Actions Projects Releases Wiki Activity

Fine-tune pack_16x16b_to_16x2b

Browse source 
Avoid mm_set1 operation when it's possible to create the constant with
one bit-shift operation from another instead. Thanks Intel for
3-operand instruction encoding!
This commit is contained in:
Pauli Oikkonen 2019-02-04 14:44:47 +02:00
parent 314f5b0e1f
commit 3f7340c932
1 changed files with 1 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
src/strategies/avx2/encode_coding_tree-avx2.c
View file

@ -71,7 +71,7 @@ static INLINE uint32_t pack_16x16b_to_16x2b(__m256i src)
* XXXX XXXX YYYY YYYY Comparison result, for movemask
*/
const __m256i threes = _mm256_set1_epi16 (3);
const __m256i cmpmask = _mm256_set1_epi16 (0x0180);
const __m256i cmpmask = _mm256_slli_epi16 (threes, 7); // 0x0180 (avoid set1)
__m256i clipped = _mm256_min_epu16 (src, threes);
__m256i shifted = _mm256_slli_epi16 (clipped, 7);