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

dec: adjust seqSymbol load on aarch64

Browse Source 
ZSTD_seqSymbol is a structure with total of 64 bits
wide. So it can be loaded in one operation and
extract its fields by simply shifting or extracting
on aarch64.
GCC doesn't recognize this and generates more
unnecessary ldr/ldrb/ldrh operations that cause
performance drop.
With this change it is observed 2~4% uplift of
silesia and 2.5~6% of cantrbry @L8 on Arm N1.

Signed-off-by: Jun He <jun.he@arm.com>
Change-Id: I7748909204cf78a17eb9d4f2333692d53239daa8
This commit is contained in:
Jun He
2022-05-30 22:01:38 +08:00
parent fda537b299
commit 2491c65937
1 changed files with 18 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
lib/decompress/zstd_decompress_block.c
+18
View File
@@ -1170,9 +1170,27 @@ FORCE_INLINE_TEMPLATE seq_t
ZSTD_decodeSequence(seqState_t* seqState, const ZSTD_longOffset_e longOffsets)
{
seq_t seq;
/*
* ZSTD_seqSymbol is a structure with a total of 64 bits wide. So it can be
* loaded in one operation and extracted its fields by simply shifting or
* bit-extracting on aarch64.
* GCC doesn't recognize this and generates more unnecessary ldr/ldrb/ldrh
* operations that cause performance drop. This can be avoided by using this
* ZSTD_memcpy hack.
*/
#if defined(__aarch64__) && (defined(__GNUC__) && !defined(__clang__))
ZSTD_seqSymbol llDInfoS, mlDInfoS, ofDInfoS;
ZSTD_seqSymbol* const llDInfo = &llDInfoS;
ZSTD_seqSymbol* const mlDInfo = &mlDInfoS;
ZSTD_seqSymbol* const ofDInfo = &ofDInfoS;
ZSTD_memcpy(llDInfo, seqState->stateLL.table + seqState->stateLL.state, sizeof(ZSTD_seqSymbol));
ZSTD_memcpy(mlDInfo, seqState->stateML.table + seqState->stateML.state, sizeof(ZSTD_seqSymbol));
ZSTD_memcpy(ofDInfo, seqState->stateOffb.table + seqState->stateOffb.state, sizeof(ZSTD_seqSymbol));
#else
const ZSTD_seqSymbol* const llDInfo = seqState->stateLL.table + seqState->stateLL.state;
const ZSTD_seqSymbol* const mlDInfo = seqState->stateML.table + seqState->stateML.state;
const ZSTD_seqSymbol* const ofDInfo = seqState->stateOffb.table + seqState->stateOffb.state;
#endif
seq.matchLength = mlDInfo->baseValue;
seq.litLength = llDInfo->baseValue;
{ U32 const ofBase = ofDInfo->baseValue;