1berry/haproxy
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

IMPORT: slz: support crc32c for lookup hash on sse4 but only if requested

Browse Source 
If building for sse4 and USE_CRC32C_HASH is defined, then we can use
crc32c to calculate the lookup hash. By default we don't do it because
even on skylake it's slower than the current hash, which only involves
a short multiply (~5% slower). But the gains are marginal (0.3%).

This is slz upstream commit 44ae4f3f85eb275adba5844d067d281e727d8850.

Note: this is not used by default and only merged in order to avoid
divergence between the code bases.
This commit is contained in:
Willy Tarreau 2023-04-09 11:35:19 +02:00
parent ea1b70900f
commit 248bbec83c
1 changed files with 5 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
src/slz.c
View File

@ -383,6 +383,11 @@ static inline uint32_t slz_hash(uint32_t a)
__asm__ volatile("crc32w %0,%0,%1" : "+r"(a) : "r"(0));
# endif
return a >> (32 - HASH_BITS);
#elif defined(__SSE4_2__) && defined(USE_CRC32C_HASH)
// SSE 4.2 offers CRC32C which is a bit slower than the multiply
// but provides a slightly smoother hash
__asm__ volatile("crc32l %1,%0" : "+r"(a) : "r"(0));
return a >> (32 - HASH_BITS);
#else
return ((a << 19) + (a << 6) - a) >> (32 - HASH_BITS);
#endif