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

Merge pull request #2040 from caoyzh/dev-2

Browse Source 
Optimize by prefetching on aarch64
This commit is contained in:
Nick Terrell
2020-04-08 13:14:47 -07:00
committed by GitHub
parent c0d4b2b5a3 7201980650
commit 5fcbc484c8
3 changed files with 11 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
lib/common/compiler.h
+3
View File
@@ -114,6 +114,9 @@
# include <mmintrin.h> /* https://msdn.microsoft.com/fr-fr/library/84szxsww(v=vs.90).aspx */ # include <mmintrin.h> /* https://msdn.microsoft.com/fr-fr/library/84szxsww(v=vs.90).aspx */
# define PREFETCH_L1(ptr) _mm_prefetch((const char*)(ptr), _MM_HINT_T0) # define PREFETCH_L1(ptr) _mm_prefetch((const char*)(ptr), _MM_HINT_T0)
# define PREFETCH_L2(ptr) _mm_prefetch((const char*)(ptr), _MM_HINT_T1) # define PREFETCH_L2(ptr) _mm_prefetch((const char*)(ptr), _MM_HINT_T1)
# elif defined(__aarch64__)
# define PREFETCH_L1(ptr) __asm__ __volatile__("prfm pldl1keep, %0" ::"Q"(*(ptr)))
# define PREFETCH_L2(ptr) __asm__ __volatile__("prfm pldl2keep, %0" ::"Q"(*(ptr)))
# elif defined(__GNUC__) && ( (__GNUC__ >= 4) || ( (__GNUC__ == 3) && (__GNUC_MINOR__ >= 1) ) ) # elif defined(__GNUC__) && ( (__GNUC__ >= 4) || ( (__GNUC__ == 3) && (__GNUC_MINOR__ >= 1) ) )
# define PREFETCH_L1(ptr) __builtin_prefetch((ptr), 0 /* rw==read */, 3 /* locality */) # define PREFETCH_L1(ptr) __builtin_prefetch((ptr), 0 /* rw==read */, 3 /* locality */)
# define PREFETCH_L2(ptr) __builtin_prefetch((ptr), 0 /* rw==read */, 2 /* locality */) # define PREFETCH_L2(ptr) __builtin_prefetch((ptr), 0 /* rw==read */, 2 /* locality */)
lib/compress/zstd_double_fast.c
+3
View File
@@ -198,6 +198,9 @@ size_t ZSTD_compressBlock_doubleFast_generic(
} } } }
ip += ((ip-anchor) >> kSearchStrength) + 1; ip += ((ip-anchor) >> kSearchStrength) + 1;
#if defined(__aarch64__)
PREFETCH_L1(ip+256);
#endif
continue; continue;
_search_next_long: _search_next_long:
lib/compress/zstd_fast.c
+5
View File
@@ -102,6 +102,11 @@ ZSTD_compressBlock_fast_generic(
const BYTE* match0 = base + matchIndex0; const BYTE* match0 = base + matchIndex0;
const BYTE* match1 = base + matchIndex1; const BYTE* match1 = base + matchIndex1;
U32 offcode; U32 offcode;
#if defined(__aarch64__)
PREFETCH_L1(ip0+256);
#endif
hashTable[h0] = current0; /* update hash table */ hashTable[h0] = current0; /* update hash table */
hashTable[h1] = current1; /* update hash table */ hashTable[h1] = current1; /* update hash table */