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

common: apply two stage copy to aarch64

Browse Source 
On aarch64 ZSTD_wildcopy uses a simple loop to do
16B based memory copy. There is existing optimized
two stage copy that can achieve better performance.
By applying this to aarch64 it is also observed ~1%
uplift in silesia corpus.

Signed-off-by: Jun He <jun.he@arm.com>
Change-Id: Ic1253308e7a8a7df2d08963ba544e086c81ce8be
This commit is contained in:
Jun He
2022-05-26 14:40:21 +08:00
parent 9a5e73c74e
commit d7249dafb4
1 changed files with 0 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
lib/common/zstd_internal.h
-7
View File
@@ -235,12 +235,6 @@ void ZSTD_wildcopy(void* dst, const void* src, ptrdiff_t length, ZSTD_overlap_e
* one COPY16() in the first call. Then, do two calls per loop since * one COPY16() in the first call. Then, do two calls per loop since
* at that point it is more likely to have a high trip count. * at that point it is more likely to have a high trip count.
*/ */
#ifdef __aarch64__
do {
COPY16(op, ip);
}
while (op < oend);
#else
ZSTD_copy16(op, ip); ZSTD_copy16(op, ip);
if (16 >= length) return; if (16 >= length) return;
op += 16; op += 16;
@@ -250,7 +244,6 @@ void ZSTD_wildcopy(void* dst, const void* src, ptrdiff_t length, ZSTD_overlap_e
COPY16(op, ip); COPY16(op, ip);
} }
while (op < oend); while (op < oend);
#endif
} }
} }