daiduo/TurboPFor-Integer-Compression
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

.

Browse Source
This commit is contained in:
powturbo
2015-05-31 21:49:54 +02:00
parent d726e7a238
commit 504931f3c4
1 changed files with 6 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
vp4dd.c
View File

@ -432,15 +432,15 @@ unsigned char *TEMPLATE2(P4DD, USIZE)(unsigned char *__restrict in, unsigned n,
uint_t *op,*pex = ex;
#if P4DN == 2
for(op = out; bb[0]; bb[0] >>= 4,op+=4) { const unsigned m = bb[0]&0xf;
_mm_storeu_si128((__m128i *)op, _mm_add_epi32(_mm_loadu_si128((__m128i*)op), _mm_shuffle_epi8(_mm_slli_epi32(_mm_loadu_si128((__m128i*)pex), b), _mm_loadu_si128((__m128i*)shuffles[m]) ) )); pex += popcnt32(m);
_mm_storeu_si128((__m128i *)op, _mm_add_epi32(_mm_loadu_si128((__m128i*)op), _mm_shuffle_epi8(_mm_slli_epi32(_mm_loadu_si128((__m128i*)pex), b), _mm_load_si128((__m128i*)shuffles[m]) ) )); pex += popcnt32(m);
}
for(op=out+64; bb[1]; bb[1] >>= 4,op+=4) { const unsigned m = bb[1]&0xf;
_mm_storeu_si128((__m128i *)op, _mm_add_epi32(_mm_loadu_si128((__m128i*)op), _mm_shuffle_epi8(_mm_slli_epi32(_mm_loadu_si128((__m128i*)pex), b), _mm_loadu_si128((__m128i*)shuffles[m]) ) )); pex += popcnt32(m);
_mm_storeu_si128((__m128i *)op, _mm_add_epi32(_mm_loadu_si128((__m128i*)op), _mm_shuffle_epi8(_mm_slli_epi32(_mm_loadu_si128((__m128i*)pex), b), _mm_load_si128((__m128i*)shuffles[m]) ) )); pex += popcnt32(m);
}
#else
for(i = 0; i < P4DN; i++) { // Loop unrolling
for(op = out; bb[i]; bb[i] >>= 4,op+=4) { const unsigned m = bb[i]&0xf;
_mm_storeu_si128((__m128i *)op, _mm_add_epi32(_mm_loadu_si128((__m128i*)op), _mm_shuffle_epi8(_mm_slli_epi32(_mm_loadu_si128((__m128i*)pex), b), _mm_loadu_si128((__m128i*)shuffles[m]) ) )); pex += popcnt32(m);
_mm_storeu_si128((__m128i *)op, _mm_add_epi32(_mm_loadu_si128((__m128i*)op), _mm_shuffle_epi8(_mm_slli_epi32(_mm_loadu_si128((__m128i*)pex), b), _mm_load_si128((__m128i*)shuffles[m]) ) )); pex += popcnt32(m);
} out+=64;
}
#endif
@ -487,15 +487,15 @@ unsigned char *TEMPLATE2(P4DDD, USIZE)(unsigned char *__restrict in, unsigned n,
uint_t *op,*pex = ex;
#if P4DN == 2
for(op = out; bb[0]; bb[0] >>= 4,op+=4) { const unsigned m = bb[0]&0xf;
_mm_storeu_si128((__m128i *)op, _mm_add_epi32(_mm_loadu_si128((__m128i*)op), _mm_shuffle_epi8(_mm_slli_epi32(_mm_loadu_si128((__m128i*)pex), b), _mm_loadu_si128((__m128i*)shuffles[m]) ) )); pex += popcnt32(m);
_mm_storeu_si128((__m128i *)op, _mm_add_epi32(_mm_loadu_si128((__m128i*)op), _mm_shuffle_epi8(_mm_slli_epi32(_mm_loadu_si128((__m128i*)pex), b), _mm_load_si128((__m128i*)shuffles[m]) ) )); pex += popcnt32(m);
}
for(op=out+64; bb[1]; bb[1] >>= 4,op+=4) { const unsigned m = bb[1]&0xf;
_mm_storeu_si128((__m128i *)op, _mm_add_epi32(_mm_loadu_si128((__m128i*)op), _mm_shuffle_epi8(_mm_slli_epi32(_mm_loadu_si128((__m128i*)pex), b), _mm_loadu_si128((__m128i*)shuffles[m]) ) )); pex += popcnt32(m);
_mm_storeu_si128((__m128i *)op, _mm_add_epi32(_mm_loadu_si128((__m128i*)op), _mm_shuffle_epi8(_mm_slli_epi32(_mm_loadu_si128((__m128i*)pex), b), _mm_load_si128((__m128i*)shuffles[m]) ) )); pex += popcnt32(m);
}
#else
for(i = 0; i < P4DN; i++) { // Loop unrolling
for(op = out; bb[i]; bb[i] >>= 4,op+=4) { const unsigned m = bb[i]&0xf;
_mm_storeu_si128((__m128i *)op, _mm_add_epi32(_mm_loadu_si128((__m128i*)op), _mm_shuffle_epi8(_mm_slli_epi32(_mm_loadu_si128((__m128i*)pex), b), _mm_loadu_si128((__m128i*)shuffles[m]) ) )); pex += popcnt32(m);
_mm_storeu_si128((__m128i *)op, _mm_add_epi32(_mm_loadu_si128((__m128i*)op), _mm_shuffle_epi8(_mm_slli_epi32(_mm_loadu_si128((__m128i*)pex), b), _mm_load_si128((__m128i*)shuffles[m]) ) )); pex += popcnt32(m);
} out+=64;
}
#endif