lib/lz4
1
0
Fork 0
mirror of https://github.com/lz4/lz4.git synced 2026-10-24 01:48:57 +03:00
Code Activity

lz4hc : copy the tables of an attached dictionary from 4 KB inputs

Browse Source 
LZ4_compress_HC_continue() copied the tables of an attached dictionary
only above 4 KB, so 4 KB pages, the most common size, used the dictCtx
search. They now use the copy.

At 4096 bytes, with a 64 KB dictionary, compared to the dictCtx search :
level 2 : -2.6% compressed size (it only searched the dictionary
when the page had no match), levels 3-9 : same output,
levels 10-11 : -0.3% / -0.1%, level 12 : same output.
Speed, AMD EPYC Genoa (Zen 4), x86-64, gcc 11 / gcc 15 / clang 22 :
level 2 : -1 to +2%, levels 3-9 : -3.3 to +4.7%, level 10 : -6 to -11%,
level 11 : +12 to +23%, level 12 : +95 to +130%.
x86 (gcc 12 -m32) : level 2 : -2.7%, levels 3-9 : -0.3 to -3.1%,
level 10 : -2.5%, level 11 : +27%, level 12 : +172%.
This commit is contained in:
Yann Collet
2026-10-22 16:10:14 -07:00
parent 7fcff05478
commit dc4983558d
2 changed files with 2 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
NEWS
View File

@@ -29,6 +29,7 @@ perf : `LZ4_FAST_DEC_LOOP` enabled on aarch64 with clang, including Android (up
perf : `LZ4_FAST_DEC_LOOP` enabled on riscv64 when unaligned accesses are fast (`-mno-strict-align`, or a `-mcpu` with fast unaligned accesses)
perf : faster compression with reused contexts, for blocks >= 1 KB : linked frames (up to +69%), `LZ4_resetStream_fast()` (up to +94%)
perf : faster compression of 2-4 KB blocks with `LZ4_attach_dictionary()` (+3% to +30% on Zen 4)
perf : 4 KB blocks with `LZ4_attach_HC_dictionary()` : level 2 compresses ~2.5% better, levels 11-12 faster (+12% to +170%), level 10 slower (-3% to -11%)
perf : level 2 compresses better (~0.2-1.5%), at similar speed (-3% to +1% on Zen 4)
perf : faster levels 11 (~+25%) and 12 (~+10%)
perf : `--favor-decSpeed` : faster decompression (~+2-4% at level 12), with fewer sequences

2
lib/lz4hc.c
View File

@@ -1483,7 +1483,7 @@ LZ4HC_compress_generic_dictCtx (
if (position >= 64 KB) {
ctx->dictCtx = NULL;
return LZ4HC_compress_generic_noDictCtx(ctx, src, dst, srcSizePtr, dstCapacity, cLevel, limit);
} else if (position == 0 && *srcSizePtr > 4 KB && isStateCompatible(ctx, ctx->dictCtx)) {
} else if (position == 0 && *srcSizePtr >= 4 KB && isStateCompatible(ctx, ctx->dictCtx)) {
/* parameters belong to the working context, not to the dictionary */
LZ4_i8 const favorDecSpeed = ctx->favorDecSpeed;
LZ4_memcpy(ctx, ctx->dictCtx, sizeof(LZ4HC_CCtx_internal));