Add and integrate lazy row hash strategy

This commit is contained in:
Nick Terrell
2021-04-07 09:53:34 -07:00
committed by Sen Huang
parent f71aabb5b5
commit 4694423c4f
16 changed files with 1256 additions and 183 deletions
+3 -1
View File
@@ -137,7 +137,8 @@ BMK_advancedParams_t BMK_initAdvancedParams(void) {
0, /* ldmHashLog */
0, /* ldmBuckSizeLog */
0, /* ldmHashRateLog */
ZSTD_lcm_auto /* literalCompressionMode */
ZSTD_lcm_auto, /* literalCompressionMode */
0 /* useRowMatchFinder */
};
return res;
}
@@ -175,6 +176,7 @@ BMK_initCCtx(ZSTD_CCtx* ctx,
CHECK_Z(ZSTD_CCtx_setParameter(ctx, ZSTD_c_nbWorkers, adv->nbWorkers));
}
CHECK_Z(ZSTD_CCtx_setParameter(ctx, ZSTD_c_compressionLevel, cLevel));
CHECK_Z(ZSTD_CCtx_setParameter(ctx, ZSTD_c_useRowMatchFinder, adv->useRowMatchFinder));
CHECK_Z(ZSTD_CCtx_setParameter(ctx, ZSTD_c_enableLongDistanceMatching, adv->ldmFlag));
CHECK_Z(ZSTD_CCtx_setParameter(ctx, ZSTD_c_ldmMinMatch, adv->ldmMinMatch));
CHECK_Z(ZSTD_CCtx_setParameter(ctx, ZSTD_c_ldmHashLog, adv->ldmHashLog));
+1
View File
@@ -117,6 +117,7 @@ typedef struct {
int ldmBucketSizeLog;
int ldmHashRateLog;
ZSTD_literalCompressionMode_e literalCompressionMode;
int useRowMatchFinder; /* use row-based matchfinder if possible */
} BMK_advancedParams_t;
/* returns default parameters used by nonAdvanced functions */
+6
View File
@@ -298,6 +298,7 @@ struct FIO_prefs_s {
int blockSize;
int overlapLog;
U32 adaptiveMode;
U32 useRowMatchFinder;
int rsyncable;
int minAdaptLevel;
int maxAdaptLevel;
@@ -468,6 +469,10 @@ void FIO_setAdaptiveMode(FIO_prefs_t* const prefs, unsigned adapt) {
prefs->adaptiveMode = adapt;
}
void FIO_setUseRowMatchFinder(FIO_prefs_t* const prefs, int useRowMatchFinder) {
prefs->useRowMatchFinder = useRowMatchFinder;
}
void FIO_setRsyncable(FIO_prefs_t* const prefs, int rsyncable) {
if ((rsyncable>0) && (prefs->nbWorkers==0))
EXM_THROW(1, "Rsyncable mode is not compatible with single thread mode \n");
@@ -986,6 +991,7 @@ static cRess_t FIO_createCResources(FIO_prefs_t* const prefs,
if (prefs->ldmHashRateLog != FIO_LDM_PARAM_NOTSET) {
CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_c_ldmHashRateLog, prefs->ldmHashRateLog) );
}
CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_c_useRowMatchFinder, prefs->useRowMatchFinder));
/* compression parameters */
CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_c_windowLog, (int)comprParams.windowLog) );
CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_c_chainLog, (int)comprParams.chainLog) );
+1
View File
@@ -77,6 +77,7 @@ void FIO_overwriteMode(FIO_prefs_t* const prefs);
void FIO_setAdaptiveMode(FIO_prefs_t* const prefs, unsigned adapt);
void FIO_setAdaptMin(FIO_prefs_t* const prefs, int minCLevel);
void FIO_setAdaptMax(FIO_prefs_t* const prefs, int maxCLevel);
void FIO_setUseRowMatchFinder(FIO_prefs_t* const prefs, int useRowMatchFinder);
void FIO_setBlockSize(FIO_prefs_t* const prefs, int blockSize);
void FIO_setChecksumFlag(FIO_prefs_t* const prefs, int checksumFlag);
void FIO_setDictIDFlag(FIO_prefs_t* const prefs, int dictIDFlag);
+7 -1
View File
@@ -205,6 +205,7 @@ static void usage_advanced(const char* programName)
DISPLAYOUT( "--long[=#]: enable long distance matching with given window log (default: %u) \n", g_defaultMaxWindowLog);
DISPLAYOUT( "--fast[=#]: switch to very fast compression levels (default: %u) \n", 1);
DISPLAYOUT( "--adapt : dynamically adapt compression level to I/O conditions \n");
DISPLAYOUT( "--[no-]row-match-finder : force enable/disable usage of fast row-based matchfinder for greedy, lazy, and lazy2 strategies \n");
# ifdef ZSTD_MULTITHREAD
DISPLAYOUT( " -T# : spawns # compression threads (default: 1, 0==# cores) \n");
DISPLAYOUT( " -B# : select size of each job (default: 0==automatic) \n");
@@ -730,6 +731,7 @@ int main(int const argCount, const char* argv[])
main_pause = 0,
nbWorkers = 0,
adapt = 0,
useRowMatchFinder = 0,
adaptMin = MINCLEVEL,
adaptMax = MAXCLEVEL,
rsyncable = 0,
@@ -857,6 +859,8 @@ int main(int const argCount, const char* argv[])
if (!strcmp(argument, "--content-size")) { contentSize = 1; continue; }
if (!strcmp(argument, "--no-content-size")) { contentSize = 0; continue; }
if (!strcmp(argument, "--adapt")) { adapt = 1; continue; }
if (!strcmp(argument, "--no-row-match-finder")) { useRowMatchFinder = 1; continue; }
if (!strcmp(argument, "--row-match-finder")) { useRowMatchFinder = 2; continue; }
if (longCommandWArg(&argument, "--adapt=")) { adapt = 1; if (!parseAdaptParameters(argument, &adaptMin, &adaptMax)) { badusage(programName); CLEAN_RETURN(1); } continue; }
if (!strcmp(argument, "--single-thread")) { nbWorkers = 0; singleThread = 1; continue; }
if (!strcmp(argument, "--format=zstd")) { suffix = ZSTD_EXTENSION; FIO_setCompressionType(prefs, FIO_zstdCompression); continue; }
@@ -1196,6 +1200,7 @@ int main(int const argCount, const char* argv[])
benchParams.ldmFlag = ldmFlag;
benchParams.ldmMinMatch = (int)g_ldmMinMatch;
benchParams.ldmHashLog = (int)g_ldmHashLog;
benchParams.useRowMatchFinder = useRowMatchFinder;
if (g_ldmBucketSizeLog != LDM_PARAM_DEFAULT) {
benchParams.ldmBucketSizeLog = (int)g_ldmBucketSizeLog;
}
@@ -1348,6 +1353,7 @@ int main(int const argCount, const char* argv[])
if (g_ldmBucketSizeLog != LDM_PARAM_DEFAULT) FIO_setLdmBucketSizeLog(prefs, (int)g_ldmBucketSizeLog);
if (g_ldmHashRateLog != LDM_PARAM_DEFAULT) FIO_setLdmHashRateLog(prefs, (int)g_ldmHashRateLog);
FIO_setAdaptiveMode(prefs, (unsigned)adapt);
FIO_setUseRowMatchFinder(prefs, useRowMatchFinder);
FIO_setAdaptMin(prefs, adaptMin);
FIO_setAdaptMax(prefs, adaptMax);
FIO_setRsyncable(prefs, rsyncable);
@@ -1387,7 +1393,7 @@ int main(int const argCount, const char* argv[])
else
operationResult = FIO_compressMultipleFilenames(fCtx, prefs, filenames->fileNames, outMirroredDirName, outDirName, outFileName, suffix, dictFileName, cLevel, compressionParams);
#else
(void)contentSize; (void)suffix; (void)adapt; (void)rsyncable; (void)ultra; (void)cLevel; (void)ldmFlag; (void)literalCompressionMode; (void)targetCBlockSize; (void)streamSrcSize; (void)srcSizeHint; (void)ZSTD_strategyMap; /* not used when ZSTD_NOCOMPRESS set */
(void)contentSize; (void)suffix; (void)adapt; (void)rsyncable; (void)ultra; (void)cLevel; (void)ldmFlag; (void)literalCompressionMode; (void)targetCBlockSize; (void)streamSrcSize; (void)srcSizeHint; (void)ZSTD_strategyMap; (void)useRowMatchFinder; /* not used when ZSTD_NOCOMPRESS set */
DISPLAY("Compression not supported \n");
#endif
} else { /* decompression or test */