Optimize ZSTD_decodeSequence by another x%
This commit is contained in:
@@ -332,7 +332,17 @@ MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getMiddleBits(size_t bitContainer, U32 c
|
||||
U32 const regMask = sizeof(bitContainer)*8 - 1;
|
||||
/* if start > regMask, bitstream is corrupted, and result is undefined */
|
||||
assert(nbBits < BIT_MASK_SIZE);
|
||||
/* x86 transform & ((1 << nbBits) - 1) to bzhi instruction, it is better
|
||||
* than accessing memory. When bmi2 instruction is not present, we consider
|
||||
* such cpus old (pre-Haswell, 2013) and their performance is not of that
|
||||
* importance.
|
||||
*/
|
||||
#if defined(__x86_64__) || defined(_M_X86)
|
||||
U64 const one = 1;
|
||||
return (bitContainer >> (start & regMask)) & ((one << nbBits) - 1);
|
||||
#else
|
||||
return (bitContainer >> (start & regMask)) & BIT_mask[nbBits];
|
||||
#endif
|
||||
}
|
||||
|
||||
MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getLowerBits(size_t bitContainer, U32 const nbBits)
|
||||
|
||||
Reference in New Issue
Block a user