cuberite/libdeflate
1
0
Fork 0
mirror of https://github.com/cuberite/libdeflate.git synced 2025-09-09 20:29:26 -04:00
Code Issues Packages Projects Releases Wiki Activity

Use shuffle+add instead of vector subscripting in SSE2 Adler-32

Browse Source 
Shuffle+add is more efficient and also works on gcc versions that don't
support subscripting of vector types (pre-4.6).
This commit is contained in:
Eric Biggers 2016-09-09 21:14:44 -07:00
parent 7d3e2a997c
commit 36f8286db7
1 changed files with 7 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
lib/adler32_impl.h
View File

@ -204,8 +204,13 @@ FUNCNAME(const void *buffer, size_t size)
(__m128i)(__v8hi){ 8, 7, 6, 5, 4, 3, 2, 1 }); (__m128i)(__v8hi){ 8, 7, 6, 5, 4, 3, 2, 1 });
/* Now accumulate what we computed into the real s1 and s2. */ /* Now accumulate what we computed into the real s1 and s2. */
s1 += v_s1[0] + v_s1[1] + v_s1[2] + v_s1[3]; v_s1 += (__v4si)_mm_shuffle_epi32((__m128i)v_s1, 0x31);
s2 += v_s2[0] + v_s2[1] + v_s2[2] + v_s2[3]; v_s1 += (__v4si)_mm_shuffle_epi32((__m128i)v_s1, 0x02);
s1 += _mm_cvtsi128_si32((__m128i)v_s1);
v_s2 += (__v4si)_mm_shuffle_epi32((__m128i)v_s2, 0x31);
v_s2 += (__v4si)_mm_shuffle_epi32((__m128i)v_s2, 0x02);
s2 += _mm_cvtsi128_si32((__m128i)v_s2);
#elif TARGET == TARGET_NEON #elif TARGET == TARGET_NEON
/* ARM NEON (Advanced SIMD) implementation */ /* ARM NEON (Advanced SIMD) implementation */