AuroraMiddleware/lz4
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

fixed clang performance in lz4_fast

Browse Source 
The simple change from
`matchIndex+MAX_DISTANCE < current`
towards
`current - matchIndex > MAX_DISTANCE`

is enough to generate a 10% performance drop under clang.
Quite massive.
(I missed as my eyes were concentrated on gcc performance at that time).

The second version is more robust, because it also survives a situation where
`matchIndex > current`
due to overflows.

The first version requires matchIndex to not overflow.
Hence were added `assert()` conditions.

The only case where this can happen is with dictCtx compression,
in the case where the dictionary context is not initialized before loading the dictionary.
So it's enough to always initialize the context while loading the dictionary.
This commit is contained in:
Yann Collet 2018-04-20 18:09:51 -07:00
parent 62d7cdcc74
commit a8a5dfd426
1 changed files with 9 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
lib/lz4.c
View File

@ -777,7 +777,8 @@ LZ4_FORCE_INLINE int LZ4_compress_generic(
LZ4_putIndexOnHash(current, h, cctx->hashTable, tableType);
if ((dictIssue == dictSmall) && (matchIndex < prefixIdxLimit)) continue; /* match outside of valid area */
if ((tableType != byU16) && (current - matchIndex > MAX_DISTANCE)) continue; /* too far - note: works even if matchIndex overflows */
assert(matchIndex < current);
if ((tableType != byU16) && (matchIndex+MAX_DISTANCE < current)) continue; /* too far - note: works even if matchIndex overflows */
if (tableType == byU16) assert((current - matchIndex) <= MAX_DISTANCE); /* too_far presumed impossible with byU16 */
if (LZ4_read32(match) == LZ4_read32(ip)) {
@ -918,8 +919,9 @@ _next_match:
match = base + matchIndex;
}
LZ4_putIndexOnHash(current, h, cctx->hashTable, tableType);
assert(matchIndex < current);
if ( ((dictIssue==dictSmall) ? (matchIndex >= prefixIdxLimit) : 1)
&& ((tableType==byU16) ? 1 : (current - matchIndex <= MAX_DISTANCE))
&& ((tableType==byU16) ? 1 : (matchIndex+MAX_DISTANCE >= current))
&& (LZ4_read32(match) == LZ4_read32(ip)) ) {
token=op++;
*token=0;
@ -1304,7 +1306,11 @@ int LZ4_loadDict (LZ4_stream_t* LZ4_dict, const char* dictionary, int dictSize)
DEBUGLOG(4, "LZ4_loadDict (%i bytes from %p into %p)", dictSize, dictionary, LZ4_dict);
LZ4_prepareTable(dict, 0, tableType);
/* It's necessary to reset the context,
* and not just continue it with prepareTable()
* to avoid any risk of generating overflowing matchIndex
* when compressing using this dictionary */
LZ4_resetStream(LZ4_dict);
/* We always increment the offset by 64 KB, since, if the dict is longer,
* we truncate it to the last 64k, and if it's shorter, we still want to