python-pillow/Pillow
1
1
Fork 0
mirror of https://github.com/python-pillow/Pillow.git synced 2025-08-21 04:34:47 +03:00
Code Issues Packages Projects Releases Wiki Activity

SIMD Filter. unroll AVX 2 times

Browse Source
This commit is contained in:
Alexander 2017-09-04 14:16:08 +03:00 committed by Alexander Karpinsky
parent 80dfbaf22d
commit 2329ee5605
1 changed files with 25 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
src/libImaging/Filter.c
View File

@ -265,7 +265,7 @@ ImagingFilter3x3(Imaging imOut, Imaging im, const float* kernel,
pix00 = _mm256_cvtepi32_ps(_mm256_cvtepu8_epi32(*(__m128i*) &in1[0])); pix00 = _mm256_cvtepi32_ps(_mm256_cvtepu8_epi32(*(__m128i*) &in1[0]));
pix10 = _mm256_cvtepi32_ps(_mm256_cvtepu8_epi32(*(__m128i*) &in0[0])); pix10 = _mm256_cvtepi32_ps(_mm256_cvtepu8_epi32(*(__m128i*) &in0[0]));
pix20 = _mm256_cvtepi32_ps(_mm256_cvtepu8_epi32(*(__m128i*) &in_1[0])); pix20 = _mm256_cvtepi32_ps(_mm256_cvtepu8_epi32(*(__m128i*) &in_1[0]));
for (; x < im->xsize-1; x += 1) { for (; x < im->xsize-1-1; x += 2) {
__m256 ss; __m256 ss;
__m128i ssi; __m128i ssi;
@ -292,6 +292,30 @@ ImagingFilter3x3(Imaging imOut, Imaging im, const float* kernel,
pix00 = _mm256_permute2f128_ps(pix00, pix01, 0x21); pix00 = _mm256_permute2f128_ps(pix00, pix01, 0x21);
pix10 = _mm256_permute2f128_ps(pix10, pix11, 0x21); pix10 = _mm256_permute2f128_ps(pix10, pix11, 0x21);
pix20 = _mm256_permute2f128_ps(pix20, pix21, 0x21); pix20 = _mm256_permute2f128_ps(pix20, pix21, 0x21);
pix01 = _mm256_permute2f128_ps(pix01, pix01, 0xf1);
pix11 = _mm256_permute2f128_ps(pix11, pix11, 0xf1);
pix21 = _mm256_permute2f128_ps(pix21, pix21, 0xf1);
ss = _mm256_set1_ps(offset);
ss = _mm256_add_ps(ss, _mm256_mul_ps(pix00, kernel00));
ss = _mm256_add_ps(ss, _mm256_mul_ps(pix01, kernel01));
ss = _mm256_add_ps(ss, _mm256_mul_ps(pix10, kernel10));
ss = _mm256_add_ps(ss, _mm256_mul_ps(pix11, kernel11));
ss = _mm256_add_ps(ss, _mm256_mul_ps(pix20, kernel20));
ss = _mm256_add_ps(ss, _mm256_mul_ps(pix21, kernel21));
ssi = _mm_cvtps_epi32(_mm_add_ps(
_mm256_extractf128_ps(ss, 0),
_mm256_extractf128_ps(ss, 1)
));
ssi = _mm_packs_epi32(ssi, ssi);
ssi = _mm_packus_epi16(ssi, ssi);
out[x+1] = _mm_cvtsi128_si32(ssi);
pix00 = _mm256_permute2f128_ps(pix00, pix01, 0x21);
pix10 = _mm256_permute2f128_ps(pix10, pix11, 0x21);
pix20 = _mm256_permute2f128_ps(pix20, pix21, 0x21);
} }
out[x] = in0[x]; out[x] = in0[x];
#else #else