mirror/opencv
1
0
Fork 0
mirror of https://github.com/opencv/opencv.git synced 2024-11-25 11:40:44 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

SymmRowSmallVec_32f 1x5 asymm

Browse Source 
NEON speedup: 2.31x
Auto-vect speedup: 2.26x

Test kernel: [-0.9432, -1.1528, 0, 1.1528, 0.9432]
This commit is contained in:
orestis 2014-12-19 22:51:42 +02:00
parent 13c0855114
commit 9c6da03504
1 changed files with 19 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
modules/imgproc/src/filter.cpp
View File

@ -3013,7 +3013,25 @@ struct SymmRowSmallVec_32f
}
else if( _ksize == 5 )
{
return 0;
float32x2_t k;
k = vdup_n_f32(0);
k = vld1_lane_f32(kx + 1, k, 0);
k = vld1_lane_f32(kx + 2, k, 1);
for( ; i <= width - 4; i += 4, src += 4 )
{
float32x4_t x0, x1, x2, x3;
x0 = vld1q_f32(src - cn);
x1 = vld1q_f32(src + cn);
x2 = vld1q_f32(src - cn*2);
x3 = vld1q_f32(src + cn*2);
float32x4_t y0;
y0 = vmulq_lane_f32(vsubq_f32(x1, x0), k, 0);
y0 = vmlaq_lane_f32(y0, vsubq_f32(x3, x2), k, 1);
vst1q_f32(dst + i, y0);
}
}
}