| /dports/graphics/vapoursynth-waifu2x-ncnn-vulkan/vapoursynth-waifu2x-ncnn-vulkan-r4/deps/ncnn/src/layer/x86/ |
| H A D | convolution_sgemm.h | 267 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
328 _sum0 = _mm256_fmadd_ps(_vb3, _va0, _sum0); // sum0 += (a30-a37) * k03 in conv_im2col_sgemm_sse()
329 _sum1 = _mm256_fmadd_ps(_vb3, _va1, _sum1); // sum1 += (a30-a37) * k13 in conv_im2col_sgemm_sse()
330 _sum2 = _mm256_fmadd_ps(_vb3, _va2, _sum2); // sum2 += (a30-a37) * k23 in conv_im2col_sgemm_sse()
536 __m256 _vb3 = _mm256_broadcast_ss(vb + 3); in conv_im2col_sgemm_sse() local
660 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
836 __m128 _vb3 = _mm_set1_ps(vb[3]); in conv_im2col_sgemm_sse() local
845 _sum3 = _mm_fmadd_ps(_va3, _vb3, _sum3); // sum3 += (k03-k33) * a30 in conv_im2col_sgemm_sse()
930 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
1404 __m128 _vb3 = _mm_set1_ps(vb[3]); in conv_im2col_sgemm_sse() local
[all …]
|
| /dports/graphics/waifu2x-ncnn-vulkan/waifu2x-ncnn-vulkan-20210521/src/ncnn/src/layer/x86/ |
| H A D | convolution_sgemm.h | 267 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
328 _sum0 = _mm256_fmadd_ps(_vb3, _va0, _sum0); // sum0 += (a30-a37) * k03 in conv_im2col_sgemm_sse()
329 _sum1 = _mm256_fmadd_ps(_vb3, _va1, _sum1); // sum1 += (a30-a37) * k13 in conv_im2col_sgemm_sse()
330 _sum2 = _mm256_fmadd_ps(_vb3, _va2, _sum2); // sum2 += (a30-a37) * k23 in conv_im2col_sgemm_sse()
536 __m256 _vb3 = _mm256_broadcast_ss(vb + 3); in conv_im2col_sgemm_sse() local
660 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
836 __m128 _vb3 = _mm_set1_ps(vb[3]); in conv_im2col_sgemm_sse() local
845 _sum3 = _mm_fmadd_ps(_va3, _vb3, _sum3); // sum3 += (k03-k33) * a30 in conv_im2col_sgemm_sse()
930 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
1404 __m128 _vb3 = _mm_set1_ps(vb[3]); in conv_im2col_sgemm_sse() local
[all …]
|
| /dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/x86/ |
| H A D | convolution_sgemm.h | 267 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
328 _sum0 = _mm256_fmadd_ps(_vb3, _va0, _sum0); // sum0 += (a30-a37) * k03 in conv_im2col_sgemm_sse()
329 _sum1 = _mm256_fmadd_ps(_vb3, _va1, _sum1); // sum1 += (a30-a37) * k13 in conv_im2col_sgemm_sse()
330 _sum2 = _mm256_fmadd_ps(_vb3, _va2, _sum2); // sum2 += (a30-a37) * k23 in conv_im2col_sgemm_sse()
536 __m256 _vb3 = _mm256_broadcast_ss(vb + 3); in conv_im2col_sgemm_sse() local
660 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
836 __m128 _vb3 = _mm_set1_ps(vb[3]); in conv_im2col_sgemm_sse() local
845 _sum3 = _mm_fmadd_ps(_va3, _vb3, _sum3); // sum3 += (k03-k33) * a30 in conv_im2col_sgemm_sse()
930 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
1404 __m128 _vb3 = _mm_set1_ps(vb[3]); in conv_im2col_sgemm_sse() local
[all …]
|
| /dports/misc/ncnn/ncnn-20211208/src/layer/x86/ |
| H A D | convolution_sgemm.h | 267 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
328 _sum0 = _mm256_comp_fmadd_ps(_vb3, _va0, _sum0); // sum0 += (a30-a37) * k03 in conv_im2col_sgemm_sse()
329 _sum1 = _mm256_comp_fmadd_ps(_vb3, _va1, _sum1); // sum1 += (a30-a37) * k13 in conv_im2col_sgemm_sse()
330 _sum2 = _mm256_comp_fmadd_ps(_vb3, _va2, _sum2); // sum2 += (a30-a37) * k23 in conv_im2col_sgemm_sse()
536 __m256 _vb3 = _mm256_broadcast_ss(vb + 3); in conv_im2col_sgemm_sse() local
660 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
836 __m128 _vb3 = _mm_set1_ps(vb[3]); in conv_im2col_sgemm_sse() local
845 _sum3 = _mm_comp_fmadd_ps(_va3, _vb3, _sum3); // sum3 += (k03-k33) * a30 in conv_im2col_sgemm_sse()
930 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
1404 __m128 _vb3 = _mm_set1_ps(vb[3]); in conv_im2col_sgemm_sse() local
[all …]
|
| /dports/graphics/realsr-ncnn-vulkan/realsr-ncnn-vulkan-20210210/src/ncnn/src/layer/x86/ |
| H A D | convolution_sgemm.h | 267 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
328 _sum0 = _mm256_fmadd_ps(_vb3, _va0, _sum0); // sum0 += (a30-a37) * k03 in conv_im2col_sgemm_sse()
329 _sum1 = _mm256_fmadd_ps(_vb3, _va1, _sum1); // sum1 += (a30-a37) * k13 in conv_im2col_sgemm_sse()
330 _sum2 = _mm256_fmadd_ps(_vb3, _va2, _sum2); // sum2 += (a30-a37) * k23 in conv_im2col_sgemm_sse()
536 __m256 _vb3 = _mm256_broadcast_ss(vb + 3); in conv_im2col_sgemm_sse() local
660 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
836 __m128 _vb3 = _mm_set1_ps(vb[3]); in conv_im2col_sgemm_sse() local
845 _sum3 = _mm_fmadd_ps(_va3, _vb3, _sum3); // sum3 += (k03-k33) * a30 in conv_im2col_sgemm_sse()
930 __m256 _vb3 = _mm256_loadu_ps(vb + 24); in conv_im2col_sgemm_sse() local
1404 __m128 _vb3 = _mm_set1_ps(vb[3]); in conv_im2col_sgemm_sse() local
[all …]
|