github_mirrors/vectorscan
3
0
Fork 1
mirror of https://github.com/VectorCamp/vectorscan.git synced 2025-06-28 16:41:01 +03:00
Code Issues Packages Projects Releases Wiki Activity

small optimization in storecompress*()

Browse Source
This commit is contained in:
Konstantinos Margaritis 2020-10-30 10:49:50 +02:00
parent 548242981d
commit 547f79b920
1 changed files with 8 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
src/util/state_compress.c
View File

@ -108,10 +108,10 @@ void storecompressed128_32bit(void *ptr, m128 xvec, m128 mvec) {
static really_inline static really_inline
void storecompressed128_64bit(void *ptr, m128 xvec, m128 mvec) { void storecompressed128_64bit(void *ptr, m128 xvec, m128 mvec) {
// First, decompose our vectors into 64-bit chunks. // First, decompose our vectors into 64-bit chunks.
u64a x[2]; u64a ALIGN_ATTR(16) x[2];
memcpy(x, &xvec, sizeof(xvec)); u64a ALIGN_ATTR(16) m[2];
u64a m[2]; store128(x, xvec);
memcpy(m, &mvec, sizeof(mvec)); store128(m, mvec);
// Count the number of bits of compressed state we're writing out per // Count the number of bits of compressed state we're writing out per
// chunk. // chunk.
@ -215,10 +215,10 @@ void storecompressed256_32bit(void *ptr, m256 xvec, m256 mvec) {
static really_really_inline static really_really_inline
void storecompressed256_64bit(void *ptr, m256 xvec, m256 mvec) { void storecompressed256_64bit(void *ptr, m256 xvec, m256 mvec) {
// First, decompose our vectors into 64-bit chunks. // First, decompose our vectors into 64-bit chunks.
u64a x[4]; u64a ALIGN_ATTR(32) x[4];
memcpy(x, &xvec, sizeof(xvec)); u64a ALIGN_ATTR(32) m[4];
u64a m[4]; store256(x, xvec);
memcpy(m, &mvec, sizeof(mvec)); store256(m, mvec);
// Count the number of bits of compressed state we're writing out per // Count the number of bits of compressed state we're writing out per
// chunk. // chunk.