Merge pull request #2733 from terrelln/huf-cspeed

[HUF] Improve Huffman encoding speed
This commit is contained in:
Nick Terrell
2021-08-03 12:59:54 -04:00
committed by GitHub
10 changed files with 539 additions and 98 deletions
+1 -1
View File
@@ -299,7 +299,7 @@ HUF_readStats_body(BYTE* huffWeight, size_t hwSize, U32* rankStats,
ZSTD_memset(rankStats, 0, (HUF_TABLELOG_MAX + 1) * sizeof(U32));
weightTotal = 0;
{ U32 n; for (n=0; n<oSize; n++) {
if (huffWeight[n] >= HUF_TABLELOG_MAX) return ERROR(corruption_detected);
if (huffWeight[n] > HUF_TABLELOG_MAX) return ERROR(corruption_detected);
rankStats[huffWeight[n]]++;
weightTotal += (1 << huffWeight[n]) >> 1;
} }
+13 -16
View File
@@ -89,9 +89,9 @@ HUF_PUBLIC_API size_t HUF_compress2 (void* dst, size_t dstCapacity,
/** HUF_compress4X_wksp() :
* Same as HUF_compress2(), but uses externally allocated `workSpace`.
* `workspace` must have minimum alignment of 4, and be at least as large as HUF_WORKSPACE_SIZE */
#define HUF_WORKSPACE_SIZE ((6 << 10) + 256)
#define HUF_WORKSPACE_SIZE_U32 (HUF_WORKSPACE_SIZE / sizeof(U32))
* `workspace` must be at least as large as HUF_WORKSPACE_SIZE */
#define HUF_WORKSPACE_SIZE ((8 << 10) + 256)
#define HUF_WORKSPACE_SIZE_U64 (HUF_WORKSPACE_SIZE / sizeof(U64))
HUF_PUBLIC_API size_t HUF_compress4X_wksp (void* dst, size_t dstCapacity,
const void* src, size_t srcSize,
unsigned maxSymbolValue, unsigned tableLog,
@@ -136,15 +136,11 @@ HUF_PUBLIC_API size_t HUF_compress4X_wksp (void* dst, size_t dstCapacity,
/* static allocation of HUF's Compression Table */
/* this is a private definition, just exposed for allocation and strict aliasing purpose. never EVER access its members directly */
struct HUF_CElt_s {
U16 val;
BYTE nbBits;
}; /* typedef'd to HUF_CElt */
typedef struct HUF_CElt_s HUF_CElt; /* consider it an incomplete type */
#define HUF_CTABLE_SIZE_U32(maxSymbolValue) ((maxSymbolValue)+1) /* Use tables of U32, for proper alignment */
#define HUF_CTABLE_SIZE(maxSymbolValue) (HUF_CTABLE_SIZE_U32(maxSymbolValue) * sizeof(U32))
typedef size_t HUF_CElt; /* consider it an incomplete type */
#define HUF_CTABLE_SIZE_ST(maxSymbolValue) ((maxSymbolValue)+2) /* Use tables of size_t, for proper alignment */
#define HUF_CTABLE_SIZE(maxSymbolValue) (HUF_CTABLE_SIZE_ST(maxSymbolValue) * sizeof(size_t))
#define HUF_CREATE_STATIC_CTABLE(name, maxSymbolValue) \
HUF_CElt name[HUF_CTABLE_SIZE_U32(maxSymbolValue)] /* no final ; */
HUF_CElt name[HUF_CTABLE_SIZE_ST(maxSymbolValue)] /* no final ; */
/* static allocation of HUF's DTable */
typedef U32 HUF_DTable;
@@ -194,6 +190,7 @@ size_t HUF_buildCTable (HUF_CElt* CTable, const unsigned* count, unsigned maxSym
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);
size_t HUF_compress4X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable);
size_t HUF_compress4X_usingCTable_bmi2(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable, int bmi2);
size_t HUF_estimateCompressedSize(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue);
int HUF_validateCTable(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue);
@@ -250,11 +247,10 @@ size_t HUF_readStats_wksp(BYTE* huffWeight, size_t hwSize,
* Loading a CTable saved with HUF_writeCTable() */
size_t HUF_readCTable (HUF_CElt* CTable, unsigned* maxSymbolValuePtr, const void* src, size_t srcSize, unsigned *hasZeroWeights);
/** HUF_getNbBits() :
/** HUF_getNbBitsFromCTable() :
* Read nbBits from CTable symbolTable, for symbol `symbolValue` presumed <= HUF_SYMBOLVALUE_MAX
* Note 1 : is not inlined, as HUF_CElt definition is private
* Note 2 : const void* used, so that it can provide a statically allocated table as argument (which uses type U32) */
U32 HUF_getNbBits(const void* symbolTable, U32 symbolValue);
* Note 1 : is not inlined, as HUF_CElt definition is private */
U32 HUF_getNbBitsFromCTable(const HUF_CElt* symbolTable, U32 symbolValue);
/*
* HUF_decompress() does the following:
@@ -306,8 +302,9 @@ size_t HUF_decompress4X2_usingDTable(void* dst, size_t maxDstSize, const void* c
/* ====================== */
size_t HUF_compress1X (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog);
size_t HUF_compress1X_wksp (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize); /**< `workSpace` must be a table of at least HUF_WORKSPACE_SIZE_U32 unsigned */
size_t HUF_compress1X_wksp (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize); /**< `workSpace` must be a table of at least HUF_WORKSPACE_SIZE_U64 U64 */
size_t HUF_compress1X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable);
size_t HUF_compress1X_usingCTable_bmi2(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable, int bmi2);
/** HUF_compress1X_repeat() :
* Same as HUF_compress1X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
* If it uses hufTable it does not modify hufTable or repeat.