Merge pull request #3285 from daniellerozenblit/optimal-huff-depth

Optimal huf depth
This commit is contained in:
daniellerozenblit
2022-10-18 10:31:44 -04:00
committed by GitHub
7 changed files with 186 additions and 121 deletions
+14 -3
View File
@@ -173,6 +173,11 @@ size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
/* ****************************************
* HUF detailed API
* ****************************************/
#define HUF_OPTIMAL_DEPTH_THRESHOLD ZSTD_btultra
typedef enum {
HUF_depth_fast, /** Use heuristic to find the table depth**/
HUF_depth_optimal /** Test possible table depths to find the one that produces the smallest header + encoded size**/
} HUF_depth_mode;
/*! HUF_compress() does the following:
* 1. count symbol occurrence from source[] into table count[] using FSE_count() (exposed within "fse.h")
@@ -185,7 +190,10 @@ size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
* For example, it's possible to compress several blocks using the same 'CTable',
* or to save and regenerate 'CTable' using external methods.
*/
unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue);
unsigned HUF_minTableLog(unsigned symbolCardinality);
unsigned HUF_cardinality(const unsigned* count, unsigned maxSymbolValue);
unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, void* workSpace,
size_t wkspSize, HUF_CElt* table, const unsigned* count, HUF_depth_mode depthMode); /* table is used as scratch space for building and testing tables, not a return value */
size_t HUF_buildCTable (HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue, unsigned maxNbBits); /* @return : maxNbBits; CTable and count can overlap. In which case, CTable will overwrite count content */
size_t HUF_writeCTable (void* dst, size_t maxDstSize, const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog);
size_t HUF_writeCTable_wksp(void* dst, size_t maxDstSize, const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog, void* workspace, size_t workspaceSize);
@@ -199,6 +207,7 @@ typedef enum {
HUF_repeat_check, /**< Can use the previous table but it must be checked. Note : The previous table must have been constructed by HUF_compress{1, 4}X_repeat */
HUF_repeat_valid /**< Can use the previous table and it is assumed to be valid */
} HUF_repeat;
/** HUF_compress4X_repeat() :
* Same as HUF_compress4X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
* If it uses hufTable it does not modify hufTable or repeat.
@@ -209,7 +218,8 @@ size_t HUF_compress4X_repeat(void* dst, size_t dstSize,
const void* src, size_t srcSize,
unsigned maxSymbolValue, unsigned tableLog,
void* workSpace, size_t wkspSize, /**< `workSpace` must be aligned on 4-bytes boundaries, `wkspSize` must be >= HUF_WORKSPACE_SIZE */
HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2, unsigned suspectUncompressible);
HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2,
unsigned suspectUncompressible, HUF_depth_mode depthMode);
/** HUF_buildCTable_wksp() :
* Same as HUF_buildCTable(), but using externally allocated scratch buffer.
@@ -315,7 +325,8 @@ size_t HUF_compress1X_repeat(void* dst, size_t dstSize,
const void* src, size_t srcSize,
unsigned maxSymbolValue, unsigned tableLog,
void* workSpace, size_t wkspSize, /**< `workSpace` must be aligned on 4-bytes boundaries, `wkspSize` must be >= HUF_WORKSPACE_SIZE */
HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2, unsigned suspectUncompressible);
HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2,
unsigned suspectUncompressible, HUF_depth_mode depthMode);
size_t HUF_decompress1X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /* single-symbol decoder */
#ifndef HUF_FORCE_DECOMPRESS_X1
+64 -17
View File
@@ -1222,15 +1222,6 @@ static size_t HUF_compressCTable_internal(
return (size_t)(op-ostart);
}
unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue)
{
unsigned tableLog = FSE_optimalTableLog_internal(maxTableLog, srcSize, maxSymbolValue, 1);
assert(tableLog <= HUF_TABLELOG_MAX);
return tableLog;
}
typedef struct {
unsigned count[HUF_SYMBOLVALUE_MAX + 1];
HUF_CElt CTable[HUF_CTABLE_SIZE_ST(HUF_SYMBOLVALUE_MAX)];
@@ -1244,6 +1235,61 @@ typedef struct {
#define SUSPECT_INCOMPRESSIBLE_SAMPLE_SIZE 4096
#define SUSPECT_INCOMPRESSIBLE_SAMPLE_RATIO 10 /* Must be >= 2 */
unsigned HUF_cardinality(const unsigned* count, unsigned maxSymbolValue)
{
unsigned cardinality = 0;
unsigned i;
for (i = 0; i < maxSymbolValue + 1; i++) {
if (count[i] != 0) cardinality += 1;
}
return cardinality;
}
unsigned HUF_minTableLog(unsigned symbolCardinality)
{
U32 minBitsSymbols = ZSTD_highbit32(symbolCardinality) + 1;
return minBitsSymbols;
}
unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, void* workSpace, size_t wkspSize, HUF_CElt* table, const unsigned* count, HUF_depth_mode depthMode)
{
unsigned optLog = FSE_optimalTableLog_internal(maxTableLog, srcSize, maxSymbolValue, 1);
assert(srcSize > 1); /* Not supported, RLE should be used instead */
if (depthMode == HUF_depth_optimal) { /** Test valid depths and return optimal **/
BYTE* dst = (BYTE*)workSpace + sizeof(HUF_WriteCTableWksp);
size_t dstSize = wkspSize - sizeof(HUF_WriteCTableWksp);
size_t optSize = ((size_t) ~0);
unsigned huffLog;
size_t maxBits, hSize, newSize;
const unsigned symbolCardinality = HUF_cardinality(count, maxSymbolValue);
if (wkspSize < sizeof(HUF_buildCTable_wksp_tables)) return optLog;
for (huffLog = HUF_minTableLog(symbolCardinality); huffLog <= maxTableLog; huffLog++) {
maxBits = HUF_buildCTable_wksp(table, count,
maxSymbolValue, huffLog,
workSpace, wkspSize);
if (ERR_isError(maxBits)) continue;
hSize = HUF_writeCTable_wksp(dst, dstSize, table, maxSymbolValue, (U32)maxBits,
workSpace, wkspSize);
if (ERR_isError(hSize)) continue;
newSize = HUF_estimateCompressedSize(table, count, maxSymbolValue) + hSize;
if (newSize < optSize) {
optSize = newSize;
optLog = huffLog;
}
}
}
assert(optLog <= HUF_TABLELOG_MAX);
return optLog;
}
/* HUF_compress_internal() :
* `workSpace_align4` must be aligned on 4-bytes boundaries,
* and occupies the same space as a table of HUF_WORKSPACE_SIZE_U64 unsigned */
@@ -1254,7 +1300,7 @@ HUF_compress_internal (void* dst, size_t dstSize,
HUF_nbStreams_e nbStreams,
void* workSpace, size_t wkspSize,
HUF_CElt* oldHufTable, HUF_repeat* repeat, int preferRepeat,
const int bmi2, unsigned suspectUncompressible)
const int bmi2, unsigned suspectUncompressible, HUF_depth_mode depthMode)
{
HUF_compress_tables_t* const table = (HUF_compress_tables_t*)HUF_alignUpWorkspace(workSpace, &wkspSize, ZSTD_ALIGNOF(size_t));
BYTE* const ostart = (BYTE*)dst;
@@ -1318,7 +1364,7 @@ HUF_compress_internal (void* dst, size_t dstSize,
}
/* Build Huffman Tree */
huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue);
huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue, &table->wksps, sizeof(table->wksps), table->CTable, table->count, depthMode);
{ size_t const maxBits = HUF_buildCTable_wksp(table->CTable, table->count,
maxSymbolValue, huffLog,
&table->wksps.buildCTable_wksp, sizeof(table->wksps.buildCTable_wksp));
@@ -1367,7 +1413,7 @@ size_t HUF_compress1X_wksp (void* dst, size_t dstSize,
return HUF_compress_internal(dst, dstSize, src, srcSize,
maxSymbolValue, huffLog, HUF_singleStream,
workSpace, wkspSize,
NULL, NULL, 0, 0 /*bmi2*/, 0);
NULL, NULL, 0, 0 /*bmi2*/, 0, HUF_depth_fast);
}
size_t HUF_compress1X_repeat (void* dst, size_t dstSize,
@@ -1375,13 +1421,13 @@ size_t HUF_compress1X_repeat (void* dst, size_t dstSize,
unsigned maxSymbolValue, unsigned huffLog,
void* workSpace, size_t wkspSize,
HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat,
int bmi2, unsigned suspectUncompressible)
int bmi2, unsigned suspectUncompressible, HUF_depth_mode depthMode)
{
DEBUGLOG(5, "HUF_compress1X_repeat (srcSize = %zu)", srcSize);
return HUF_compress_internal(dst, dstSize, src, srcSize,
maxSymbolValue, huffLog, HUF_singleStream,
workSpace, wkspSize, hufTable,
repeat, preferRepeat, bmi2, suspectUncompressible);
repeat, preferRepeat, bmi2, suspectUncompressible, depthMode);
}
/* HUF_compress4X_repeat():
@@ -1396,7 +1442,7 @@ size_t HUF_compress4X_wksp (void* dst, size_t dstSize,
return HUF_compress_internal(dst, dstSize, src, srcSize,
maxSymbolValue, huffLog, HUF_fourStreams,
workSpace, wkspSize,
NULL, NULL, 0, 0 /*bmi2*/, 0);
NULL, NULL, 0, 0 /*bmi2*/, 0, HUF_depth_fast);
}
/* HUF_compress4X_repeat():
@@ -1407,13 +1453,14 @@ size_t HUF_compress4X_repeat (void* dst, size_t dstSize,
const void* src, size_t srcSize,
unsigned maxSymbolValue, unsigned huffLog,
void* workSpace, size_t wkspSize,
HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2, unsigned suspectUncompressible)
HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2,
unsigned suspectUncompressible, HUF_depth_mode depthMode)
{
DEBUGLOG(5, "HUF_compress4X_repeat (srcSize = %zu)", srcSize);
return HUF_compress_internal(dst, dstSize, src, srcSize,
maxSymbolValue, huffLog, HUF_fourStreams,
workSpace, wkspSize,
hufTable, repeat, preferRepeat, bmi2, suspectUncompressible);
hufTable, repeat, preferRepeat, bmi2, suspectUncompressible, depthMode);
}
#ifndef ZSTD_NO_UNUSED_FUNCTIONS
+9 -4
View File
@@ -2653,6 +2653,8 @@ ZSTD_entropyCompressSeqStore_internal(seqStore_t* seqStorePtr,
/* Base suspicion of uncompressibility on ratio of literals to sequences */
unsigned const suspectUncompressible = (numSequences == 0) || (numLiterals / numSequences >= SUSPECT_UNCOMPRESSIBLE_LITERAL_RATIO);
size_t const litSize = (size_t)(seqStorePtr->lit - literals);
HUF_depth_mode depthMode = cctxParams->cParams.strategy >= HUF_OPTIMAL_DEPTH_THRESHOLD ? HUF_depth_optimal : HUF_depth_fast;
size_t const cSize = ZSTD_compressLiterals(
&prevEntropy->huf, &nextEntropy->huf,
cctxParams->cParams.strategy,
@@ -2660,7 +2662,7 @@ ZSTD_entropyCompressSeqStore_internal(seqStore_t* seqStorePtr,
op, dstCapacity,
literals, litSize,
entropyWorkspace, entropyWkspSize,
bmi2, suspectUncompressible);
bmi2, suspectUncompressible, depthMode);
FORWARD_IF_ERROR(cSize, "ZSTD_compressLiterals failed");
assert(cSize <= dstCapacity);
op += cSize;
@@ -3107,7 +3109,7 @@ static size_t ZSTD_buildBlockEntropyStats_literals(void* const src, size_t srcSi
ZSTD_hufCTables_t* nextHuf,
ZSTD_hufCTablesMetadata_t* hufMetadata,
const int literalsCompressionIsDisabled,
void* workspace, size_t wkspSize)
void* workspace, size_t wkspSize, HUF_depth_mode depthMode)
{
BYTE* const wkspStart = (BYTE*)workspace;
BYTE* const wkspEnd = wkspStart + wkspSize;
@@ -3164,7 +3166,7 @@ static size_t ZSTD_buildBlockEntropyStats_literals(void* const src, size_t srcSi
/* Build Huffman Tree */
ZSTD_memset(nextHuf->CTable, 0, sizeof(nextHuf->CTable));
huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue);
huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue, nodeWksp, nodeWkspSize, nextHuf->CTable, countWksp, depthMode);
assert(huffLog <= LitHufLog);
{ size_t const maxBits = HUF_buildCTable_wksp((HUF_CElt*)nextHuf->CTable, countWksp,
maxSymbolValue, huffLog,
@@ -3268,12 +3270,15 @@ size_t ZSTD_buildBlockEntropyStats(seqStore_t* seqStorePtr,
void* workspace, size_t wkspSize)
{
size_t const litSize = seqStorePtr->lit - seqStorePtr->litStart;
HUF_depth_mode depthMode = cctxParams->cParams.strategy >= HUF_OPTIMAL_DEPTH_THRESHOLD ? HUF_depth_optimal : HUF_depth_fast;
entropyMetadata->hufMetadata.hufDesSize =
ZSTD_buildBlockEntropyStats_literals(seqStorePtr->litStart, litSize,
&prevEntropy->huf, &nextEntropy->huf,
&entropyMetadata->hufMetadata,
ZSTD_literalsCompressionIsDisabled(cctxParams),
workspace, wkspSize);
workspace, wkspSize, depthMode);
FORWARD_IF_ERROR(entropyMetadata->hufMetadata.hufDesSize, "ZSTD_buildBlockEntropyStats_literals failed");
entropyMetadata->fseMetadata.fseTablesSize =
ZSTD_buildBlockEntropyStats_sequences(seqStorePtr,
+3 -3
View File
@@ -99,7 +99,7 @@ size_t ZSTD_compressLiterals (ZSTD_hufCTables_t const* prevHuf,
const void* src, size_t srcSize,
void* entropyWorkspace, size_t entropyWorkspaceSize,
const int bmi2,
unsigned suspectUncompressible)
unsigned suspectUncompressible, HUF_depth_mode depthMode)
{
size_t const minGain = ZSTD_minGain(srcSize, strategy);
size_t const lhSize = 3 + (srcSize >= 1 KB) + (srcSize >= 16 KB);
@@ -128,7 +128,7 @@ size_t ZSTD_compressLiterals (ZSTD_hufCTables_t const* prevHuf,
RETURN_ERROR_IF(dstCapacity < lhSize+1, dstSize_tooSmall, "not enough space for compression");
{ HUF_repeat repeat = prevHuf->repeatMode;
int const preferRepeat = (strategy < ZSTD_lazy) ? srcSize <= 1024 : 0;
typedef size_t (*huf_compress_f)(void*, size_t, const void*, size_t, unsigned, unsigned, void*, size_t, HUF_CElt*, HUF_repeat*, int, int, unsigned);
typedef size_t (*huf_compress_f)(void*, size_t, const void*, size_t, unsigned, unsigned, void*, size_t, HUF_CElt*, HUF_repeat*, int, int, unsigned, HUF_depth_mode);
huf_compress_f huf_compress;
if (repeat == HUF_repeat_valid && lhSize == 3) singleStream = 1;
huf_compress = singleStream ? HUF_compress1X_repeat : HUF_compress4X_repeat;
@@ -138,7 +138,7 @@ size_t ZSTD_compressLiterals (ZSTD_hufCTables_t const* prevHuf,
entropyWorkspace, entropyWorkspaceSize,
(HUF_CElt*)nextHuf->CTable,
&repeat, preferRepeat,
bmi2, suspectUncompressible);
bmi2, suspectUncompressible, depthMode);
if (repeat != HUF_repeat_none) {
/* reused the existing table */
DEBUGLOG(5, "Reusing previous huffman table");
+1 -1
View File
@@ -26,6 +26,6 @@ size_t ZSTD_compressLiterals (ZSTD_hufCTables_t const* prevHuf,
const void* src, size_t srcSize,
void* entropyWorkspace, size_t entropyWorkspaceSize,
const int bmi2,
unsigned suspectUncompressible);
unsigned suspectUncompressible, HUF_depth_mode depthMode);
#endif /* ZSTD_COMPRESS_LITERALS_H */