sharpetronics/zstd
1
0
Fork 0
mirror of https://github.com/facebook/zstd.git synced 2025-10-04 00:02:33 -04:00
Code Issues Projects Releases Wiki Activity

Alignment instruction ZSTD_ALIGNED() in common/compiler.h

Browse Source
This commit is contained in:
Yann Collet 2025-01-14 15:54:10 -08:00 committed by Yann Collet
parent 2f3ee8b530
commit 8bff69af86
2 changed files with 15 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
lib/common/compiler.h
View File

@ -281,7 +281,7 @@
#endif
/*-**************************************************************
* Alignment check
* Alignment
*****************************************************************/
/* @return 1 if @u is a 2^n value, 0 otherwise
@ -315,6 +315,19 @@ MEM_STATIC int ZSTD_isPower2(size_t u) {
# endif
#endif /* ZSTD_ALIGNOF */
#ifndef ZSTD_ALIGNED
/* C90-compatible alignment macro (GCC/Clang). Adjust for other compilers if needed. */
# if defined(__GNUC__)
# define ZSTD_ALIGNED(a) __attribute__((aligned(a)))
# elif defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L) /* C11 */
# define ZSTD_ALIGNED(a) alignas(a)
# else
/* this compiler will require its own alignment instruction */
# define ZSTD_ALIGNED(...)
# endif
#endif /* ZSTD_ALIGNED */
/*-**************************************************************
* Sanitizer
*****************************************************************/

12
lib/compress/zstd_compress.c
View File

@ -7386,22 +7386,12 @@ size_t ZSTD_convertBlockSequences(ZSTD_CCtx* cctx,
#if defined(ZSTD_ARCH_X86_AVX2)
/* C90-compatible alignment macro (GCC/Clang). Adjust for other compilers if needed. */
#if defined(__GNUC__)
# define ALIGNED32 __attribute__((aligned(32)))
#elif defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L) /* C11 */
# define ALIGNED32 alignas(32)
#else
/* this compiler will require its own alignment instruction */
# define ALIGNED32
#endif
BlockSummary ZSTD_get1BlockSummary(const ZSTD_Sequence* seqs, size_t nbSeqs)
{
size_t i;
__m256i const zeroVec = _mm256_setzero_si256();
__m256i sumVec = zeroVec; /* accumulates match+lit in 32-bit lanes */
ALIGNED32 U32 tmp[8]; /* temporary buffer for reduction */
ZSTD_ALIGNED(32) U32 tmp[8]; /* temporary buffer for reduction */
size_t mSum = 0, lSum = 0;
ZSTD_STATIC_ASSERT(sizeof(ZSTD_Sequence) == 16);