ddidderr/zstd-rs
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Reduce RowHash's tag space size by x2 (#3543)

Browse Source 
Allocate half the memory for tag space, which means that we get one less slot for an actual tag (needs to be used for next position index).
The results is a slight loss in compression ratio (up to 0.2%) and some regressions/improvements to speed depending on level and sample. In turn, we get to save 16% of the hash table's space (5 bytes per entry instead of 6 bytes per entry).
This commit is contained in:
Yonatan Komornik
2023-03-10 14:15:04 -08:00
committed by GitHub
parent 134d332b10
commit 33e39094e7
5 changed files with 320 additions and 321 deletions
Download Patch File Download Diff File Expand all files Collapse all files
lib/compress/zstd_compress.c
+5 -5
View File
@@ -1616,7 +1616,7 @@ ZSTD_sizeof_matchState(const ZSTD_compressionParameters* const cParams,
+ ZSTD_cwksp_aligned_alloc_size((ZSTD_OPT_NUM+1) * sizeof(ZSTD_match_t))
+ ZSTD_cwksp_aligned_alloc_size((ZSTD_OPT_NUM+1) * sizeof(ZSTD_optimal_t));
size_t const lazyAdditionalSpace = ZSTD_rowMatchFinderUsed(cParams->strategy, useRowMatchFinder)
? ZSTD_cwksp_aligned_alloc_size(hSize*sizeof(U16))
? ZSTD_cwksp_aligned_alloc_size(hSize)
: 0;
size_t const optSpace = (forCCtx && (cParams->strategy >= ZSTD_btopt))
? optPotentialSpace
@@ -1968,8 +1968,8 @@ ZSTD_reset_matchState(ZSTD_matchState_t* ms,
if (ZSTD_rowMatchFinderUsed(cParams->strategy, useRowMatchFinder)) {
{ /* Row match finder needs an additional table of hashes ("tags") */
size_t const tagTableSize = hSize*sizeof(U16);
ms->tagTable = (U16*)ZSTD_cwksp_reserve_aligned(ws, tagTableSize);
size_t const tagTableSize = hSize;
ms->tagTable = (BYTE*)ZSTD_cwksp_reserve_aligned(ws, tagTableSize);
if (ms->tagTable) ZSTD_memset(ms->tagTable, 0, tagTableSize);
}
{ /* Switch to 32-entry rows if searchLog is 5 (or more) */
@@ -2362,7 +2362,7 @@ static size_t ZSTD_resetCCtx_byCopyingCDict(ZSTD_CCtx* cctx,
}
/* copy tag table */
if (ZSTD_rowMatchFinderUsed(cdict_cParams->strategy, cdict->useRowMatchFinder)) {
size_t const tagTableSize = hSize*sizeof(U16);
size_t const tagTableSize = hSize;
ZSTD_memcpy(cctx->blockState.matchState.tagTable,
cdict->matchState.tagTable,
tagTableSize);
@@ -4713,7 +4713,7 @@ static size_t ZSTD_loadDictionaryContent(ZSTD_matchState_t* ms,
} else {
assert(params->useRowMatchFinder != ZSTD_ps_auto);
if (params->useRowMatchFinder == ZSTD_ps_enable) {
size_t const tagTableSize = ((size_t)1 << params->cParams.hashLog) * sizeof(U16);
size_t const tagTableSize = ((size_t)1 << params->cParams.hashLog);
ZSTD_memset(ms->tagTable, 0, tagTableSize);
ZSTD_row_update(ms, iend-HASH_READ_SIZE);
DEBUGLOG(4, "Using row-based hash table for lazy dict");