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

Move bitwise builtins into bits.h

Browse Source
This commit is contained in:
Elliot Gorokhovsky
2022-02-14 11:16:03 -05:00
parent 970460f67d
commit db2f4a6532
16 changed files with 250 additions and 400 deletions
Download Patch File Download Diff File Expand all files Collapse all files
lib/compress/fse_compress.c
+5 -4
View File
@@ -26,6 +26,7 @@
#define ZSTD_DEPS_NEED_MALLOC
#define ZSTD_DEPS_NEED_MATH64
#include "../common/zstd_deps.h" /* ZSTD_malloc, ZSTD_free, ZSTD_memcpy, ZSTD_memset */
#include "../common/bits.h" /* ZSTD_highbit32 */
/* **************************************************************
@@ -191,7 +192,7 @@ size_t FSE_buildCTable_wksp(FSE_CTable* ct,
break;
default :
assert(normalizedCounter[s] > 1);
{ U32 const maxBitsOut = tableLog - BIT_highbit32 ((U32)normalizedCounter[s]-1);
{ U32 const maxBitsOut = tableLog - ZSTD_highbit32 ((U32)normalizedCounter[s]-1);
U32 const minStatePlus = (U32)normalizedCounter[s] << maxBitsOut;
symbolTT[s].deltaNbBits = (maxBitsOut << 16) - minStatePlus;
symbolTT[s].deltaFindState = (int)(total - (unsigned)normalizedCounter[s]);
@@ -355,8 +356,8 @@ void FSE_freeCTable (FSE_CTable* ct) { ZSTD_free(ct); }
/* provides the minimum logSize to safely represent a distribution */
static unsigned FSE_minTableLog(size_t srcSize, unsigned maxSymbolValue)
{
U32 minBitsSrc = BIT_highbit32((U32)(srcSize)) + 1;
U32 minBitsSymbols = BIT_highbit32(maxSymbolValue) + 2;
U32 minBitsSrc = ZSTD_highbit32((U32)(srcSize)) + 1;
U32 minBitsSymbols = ZSTD_highbit32(maxSymbolValue) + 2;
U32 minBits = minBitsSrc < minBitsSymbols ? minBitsSrc : minBitsSymbols;
assert(srcSize > 1); /* Not supported, RLE should be used instead */
return minBits;
@@ -364,7 +365,7 @@ static unsigned FSE_minTableLog(size_t srcSize, unsigned maxSymbolValue)
unsigned FSE_optimalTableLog_internal(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, unsigned minus)
{
U32 maxBitsSrc = BIT_highbit32((U32)(srcSize - 1)) - minus;
U32 maxBitsSrc = ZSTD_highbit32((U32)(srcSize - 1)) - minus;
U32 tableLog = maxTableLog;
U32 minBits = FSE_minTableLog(srcSize, maxSymbolValue);
assert(srcSize > 1); /* Not supported, RLE should be used instead */
lib/compress/huf_compress.c
+5 -4
View File
@@ -32,6 +32,7 @@
#define HUF_STATIC_LINKING_ONLY
#include "../common/huf.h"
#include "../common/error_private.h"
#include "../common/bits.h" /* ZSTD_highbit32 */
/* **************************************************************
@@ -407,7 +408,7 @@ static U32 HUF_setMaxHeight(nodeElt* huffNode, U32 lastNonNull, U32 targetNbBits
/* Try to reduce the next power of 2 above totalCost because we
* gain back half the rank.
*/
U32 nBitsToDecrease = BIT_highbit32((U32)totalCost) + 1;
U32 nBitsToDecrease = ZSTD_highbit32((U32)totalCost) + 1;
for ( ; nBitsToDecrease > 1; nBitsToDecrease--) {
U32 const highPos = rankLast[nBitsToDecrease];
U32 const lowPos = rankLast[nBitsToDecrease-1];
@@ -505,7 +506,7 @@ typedef struct {
*/
#define RANK_POSITION_MAX_COUNT_LOG 32
#define RANK_POSITION_LOG_BUCKETS_BEGIN (RANK_POSITION_TABLE_SIZE - 1) - RANK_POSITION_MAX_COUNT_LOG - 1 /* == 158 */
#define RANK_POSITION_DISTINCT_COUNT_CUTOFF RANK_POSITION_LOG_BUCKETS_BEGIN + BIT_highbit32(RANK_POSITION_LOG_BUCKETS_BEGIN) /* == 166 */
#define RANK_POSITION_DISTINCT_COUNT_CUTOFF RANK_POSITION_LOG_BUCKETS_BEGIN + ZSTD_highbit32(RANK_POSITION_LOG_BUCKETS_BEGIN) /* == 166 */
/* Return the appropriate bucket index for a given count. See definition of
* RANK_POSITION_DISTINCT_COUNT_CUTOFF for explanation of bucketing strategy.
@@ -513,7 +514,7 @@ typedef struct {
static U32 HUF_getIndex(U32 const count) {
return (count < RANK_POSITION_DISTINCT_COUNT_CUTOFF)
? count
: BIT_highbit32(count) + RANK_POSITION_LOG_BUCKETS_BEGIN;
: ZSTD_highbit32(count) + RANK_POSITION_LOG_BUCKETS_BEGIN;
}
/* Helper swap function for HUF_quickSortPartition() */
@@ -870,7 +871,7 @@ FORCE_INLINE_TEMPLATE void HUF_addBits(HUF_CStream_t* bitC, HUF_CElt elt, int id
#if DEBUGLEVEL >= 1
{
size_t const nbBits = HUF_getNbBits(elt);
size_t const dirtyBits = nbBits == 0 ? 0 : BIT_highbit32((U32)nbBits) + 1;
size_t const dirtyBits = nbBits == 0 ? 0 : ZSTD_highbit32((U32)nbBits) + 1;
(void)dirtyBits;
/* Middle bits are 0. */
assert(((elt >> dirtyBits) << (dirtyBits + nbBits)) == 0);
lib/compress/zstd_compress.c
+1
View File
@@ -27,6 +27,7 @@
#include "zstd_opt.h"
#include "zstd_ldm.h"
#include "zstd_compress_superblock.h"
#include "../common/bits.h" /* ZSTD_highbit32 */
/* ***************************************************************
* Tuning parameters
lib/compress/zstd_compress_internal.h
+1 -97
View File
@@ -23,6 +23,7 @@
#ifdef ZSTD_MULTITHREAD
# include "zstdmt_compress.h"
#endif
#include "../common/bits.h" /* ZSTD_highbit32, ZSTD_NbCommonBytes */
#if defined (__cplusplus)
extern "C" {
@@ -699,103 +700,6 @@ ZSTD_newRep(U32 const rep[ZSTD_REP_NUM], U32 const offBase, U32 const ll0)
/*-*************************************
* Match length counter
***************************************/
static unsigned ZSTD_NbCommonBytes (size_t val)
{
if (MEM_isLittleEndian()) {
if (MEM_64bits()) {
# if defined(_MSC_VER) && defined(_WIN64)
# if STATIC_BMI2
return _tzcnt_u64(val) >> 3;
# else
if (val != 0) {
unsigned long r;
_BitScanForward64(&r, (U64)val);
return (unsigned)(r >> 3);
} else {
/* Should not reach this code path */
__assume(0);
}
# endif
# elif defined(__GNUC__) && (__GNUC__ >= 4)
return (unsigned)(__builtin_ctzll((U64)val) >> 3);
# else
static const int DeBruijnBytePos[64] = { 0, 0, 0, 0, 0, 1, 1, 2,
0, 3, 1, 3, 1, 4, 2, 7,
0, 2, 3, 6, 1, 5, 3, 5,
1, 3, 4, 4, 2, 5, 6, 7,
7, 0, 1, 2, 3, 3, 4, 6,
2, 6, 5, 5, 3, 4, 5, 6,
7, 1, 2, 4, 6, 4, 4, 5,
7, 2, 6, 5, 7, 6, 7, 7 };
return DeBruijnBytePos[((U64)((val & -(long long)val) * 0x0218A392CDABBD3FULL)) >> 58];
# endif
} else { /* 32 bits */
# if defined(_MSC_VER)
if (val != 0) {
unsigned long r;
_BitScanForward(&r, (U32)val);
return (unsigned)(r >> 3);
} else {
/* Should not reach this code path */
__assume(0);
}
# elif defined(__GNUC__) && (__GNUC__ >= 3)
return (unsigned)(__builtin_ctz((U32)val) >> 3);
# else
static const int DeBruijnBytePos[32] = { 0, 0, 3, 0, 3, 1, 3, 0,
3, 2, 2, 1, 3, 2, 0, 1,
3, 3, 1, 2, 2, 2, 2, 0,
3, 1, 2, 0, 1, 0, 1, 1 };
return DeBruijnBytePos[((U32)((val & -(S32)val) * 0x077CB531U)) >> 27];
# endif
}
} else { /* Big Endian CPU */
if (MEM_64bits()) {
# if defined(_MSC_VER) && defined(_WIN64)
# if STATIC_BMI2
return _lzcnt_u64(val) >> 3;
# else
if (val != 0) {
unsigned long r;
_BitScanReverse64(&r, (U64)val);
return (unsigned)(r >> 3);
} else {
/* Should not reach this code path */
__assume(0);
}
# endif
# elif defined(__GNUC__) && (__GNUC__ >= 4)
return (unsigned)(__builtin_clzll(val) >> 3);
# else
unsigned r;
const unsigned n32 = sizeof(size_t)*4; /* calculate this way due to compiler complaining in 32-bits mode */
if (!(val>>n32)) { r=4; } else { r=0; val>>=n32; }
if (!(val>>16)) { r+=2; val>>=8; } else { val>>=24; }
r += (!val);
return r;
# endif
} else { /* 32 bits */
# if defined(_MSC_VER)
if (val != 0) {
unsigned long r;
_BitScanReverse(&r, (unsigned long)val);
return (unsigned)(r >> 3);
} else {
/* Should not reach this code path */
__assume(0);
}
# elif defined(__GNUC__) && (__GNUC__ >= 3)
return (unsigned)(__builtin_clz((U32)val) >> 3);
# else
unsigned r;
if (!(val>>16)) { r=2; val>>=8; } else { r=0; val>>=24; }
r += (!val);
return r;
# endif
} }
}
MEM_STATIC size_t ZSTD_count(const BYTE* pIn, const BYTE* pMatch, const BYTE* const pInLimit)
{
const BYTE* const pStart = pIn;
lib/compress/zstd_lazy.c
+3 -40
View File
@@ -10,6 +10,7 @@
#include "zstd_compress_internal.h"
#include "zstd_lazy.h"
#include "../common/bits.h" /* ZSTD_countTrailingZeros64 */
/*-*************************************
@@ -765,44 +766,6 @@ size_t ZSTD_HcFindBestMatch(
typedef U64 ZSTD_VecMask; /* Clarifies when we are interacting with a U64 representing a mask of matches */
/* ZSTD_VecMask_next():
* Starting from the LSB, returns the idx of the next non-zero bit.
* Basically counting the nb of trailing zeroes.
*/
static U32 ZSTD_VecMask_next(ZSTD_VecMask val) {
assert(val != 0);
# if defined(_MSC_VER) && defined(_WIN64)
if (val != 0) {
unsigned long r;
_BitScanForward64(&r, val);
return (U32)(r);
} else {
/* Should not reach this code path */
__assume(0);
}
# elif (defined(__GNUC__) && ((__GNUC__ > 3) || ((__GNUC__ == 3) && (__GNUC_MINOR__ >= 4))))
if (sizeof(size_t) == 4) {
U32 mostSignificantWord = (U32)(val >> 32);
U32 leastSignificantWord = (U32)val;
if (leastSignificantWord == 0) {
return 32 + (U32)__builtin_ctz(mostSignificantWord);
} else {
return (U32)__builtin_ctz(leastSignificantWord);
}
} else {
return (U32)__builtin_ctzll(val);
}
# else
/* Software ctz version: http://aggregate.org/MAGIC/#Trailing%20Zero%20Count
* and: https://stackoverflow.com/questions/2709430/count-number-of-bits-in-a-64-bit-long-big-integer
*/
val = ~val & (val - 1ULL); /* Lowest set bit mask */
val = val - ((val >> 1) & 0x5555555555555555);
val = (val & 0x3333333333333333ULL) + ((val >> 2) & 0x3333333333333333ULL);
return (U32)((((val + (val >> 4)) & 0xF0F0F0F0F0F0F0FULL) * 0x101010101010101ULL) >> 56);
# endif
}
/* ZSTD_rotateRight_*():
* Rotates a bitfield to the right by "count" bits.
* https://en.wikipedia.org/w/index.php?title=Circular_shift&oldid=991635599#Implementing_circular_shifts
@@ -1202,7 +1165,7 @@ size_t ZSTD_RowFindBestMatch(
/* Cycle through the matches and prefetch */
for (; (matches > 0) && (nbAttempts > 0); --nbAttempts, matches &= (matches - 1)) {
U32 const matchPos = (head + ZSTD_VecMask_next(matches)) & rowMask;
U32 const matchPos = (head + ZSTD_countTrailingZeros64(matches)) & rowMask;
U32 const matchIndex = row[matchPos];
assert(numMatches < rowEntries);
if (matchIndex < lowLimit)
@@ -1270,7 +1233,7 @@ size_t ZSTD_RowFindBestMatch(
ZSTD_VecMask matches = ZSTD_row_getMatchMask(dmsTagRow, (BYTE)dmsTag, head, rowEntries);
for (; (matches > 0) && (nbAttempts > 0); --nbAttempts, matches &= (matches - 1)) {
U32 const matchPos = (head + ZSTD_VecMask_next(matches)) & rowMask;
U32 const matchPos = (head + ZSTD_countTrailingZeros64(matches)) & rowMask;
U32 const matchIndex = dmsRow[matchPos];
if (matchIndex < dmsLowestIndex)
break;