generic-library/opencv
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Fix SymmColumnSmallVec_32s16s SSE code

Browse Source 
Specifically, in general antisymmetric case src[-1] (S0) should be
subtracted from from src[1] (S2), and not the opposite.
This commit is contained in:
orestis 2014-12-06 19:53:07 +02:00
parent 0a4c616d43
commit c762da9893
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
modules/imgproc/src/filter.cpp
View File

@ -1182,10 +1182,10 @@ struct SymmColumnSmallVec_32s16s
{
__m128 s0 = df4, s1 = df4;
__m128i x0, x1;
x0 = _mm_sub_epi32(_mm_load_si128((__m128i*)(S0 + i)),
_mm_load_si128((__m128i*)(S2 + i)));
x1 = _mm_sub_epi32(_mm_load_si128((__m128i*)(S0 + i + 4)),
_mm_load_si128((__m128i*)(S2 + i + 4)));
x0 = _mm_sub_epi32(_mm_load_si128((__m128i*)(S2 + i)),
_mm_load_si128((__m128i*)(S0 + i)));
x1 = _mm_sub_epi32(_mm_load_si128((__m128i*)(S2 + i + 4)),
_mm_load_si128((__m128i*)(S0 + i + 4)));
s0 = _mm_add_ps(s0, _mm_mul_ps(_mm_cvtepi32_ps(x0),k1));
s1 = _mm_add_ps(s1, _mm_mul_ps(_mm_cvtepi32_ps(x1),k1));
x0 = _mm_packs_epi32(_mm_cvtps_epi32(s0), _mm_cvtps_epi32(s1));