mirrors/cosmopolitan
1
0
Fork 0
mirror of https://github.com/jart/cosmopolitan.git synced 2025-10-09 06:08:23 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

Support avx512f + vpclmulqdq crc32() acceleration

Browse source 
Cosmo's _Cz_crc32() function now goes 73 GiB/s on Threadripper. This
will significantly improve the performance of the PKZIP file format.
This algorithm is also used by apelink, to create deterministic ids.
This commit is contained in:
Justine Tunney 2024-05-29 10:13:37 -07:00
parent 7c8df05042
commit a05ce3ad9d
No known key found for this signature in database
GPG key ID: BE714B4575D6E328
8 changed files with 385 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

5
third_party/zlib/crc32.c vendored
View file

@ -780,6 +780,7 @@ uint32_t ZEXPORT crc32_z(crc, buf_, len)
}
#endif
#if defined(__x86_64__)
#if defined(CRC32_SIMD_AVX512_PCLMUL)
if (x86_cpu_enable_avx512 && len >= Z_CRC32_AVX512_MINIMUM_LENGTH) {
/* crc32 64-byte chunks */
@ -792,7 +793,8 @@ uint32_t ZEXPORT crc32_z(crc, buf_, len)
/* Fall into the default crc32 for the remaining data. */
buf += chunk_size;
}
#elif defined(CRC32_SIMD_SSE42_PCLMUL)
#endif
#if defined(CRC32_SIMD_SSE42_PCLMUL)
if (x86_cpu_enable_simd && len >= Z_CRC32_SSE42_MINIMUM_LENGTH) {
/* crc32 16-byte chunks */
z_size_t chunk_size = len & ~Z_CRC32_SSE42_CHUNKSIZE_MASK;
@ -804,6 +806,7 @@ uint32_t ZEXPORT crc32_z(crc, buf_, len)
/* Fall into the default crc32 for the remaining data. */
buf += chunk_size;
}
#endif
#elif defined(CRC32_ARMV8_CRC32)
if (arm_cpu_enable_crc32) {
#if defined(__aarch64__)