| /dports/graphics/vapoursynth-waifu2x-ncnn-vulkan/vapoursynth-waifu2x-ncnn-vulkan-r4/deps/ncnn/src/layer/x86/ |
| H A D | convolution_sgemm.h | 266 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
308 _sum0 = _mm256_fmadd_ps(_vb2, _va0, _sum0); // sum0 += (a20-a27) * k02 in conv_im2col_sgemm_sse()
309 _sum1 = _mm256_fmadd_ps(_vb2, _va1, _sum1); // sum1 += (a20-a27) * k12 in conv_im2col_sgemm_sse()
310 _sum2 = _mm256_fmadd_ps(_vb2, _va2, _sum2); // sum2 += (a20-a27) * k22 in conv_im2col_sgemm_sse()
535 __m256 _vb2 = _mm256_broadcast_ss(vb + 2); in conv_im2col_sgemm_sse() local
659 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
835 __m128 _vb2 = _mm_set1_ps(vb[2]); in conv_im2col_sgemm_sse() local
844 _sum2 = _mm_fmadd_ps(_va2, _vb2, _sum2); // sum2 += (k02-k32) * a20 in conv_im2col_sgemm_sse()
929 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
1403 __m128 _vb2 = _mm_set1_ps(vb[2]); 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 | 266 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
308 _sum0 = _mm256_fmadd_ps(_vb2, _va0, _sum0); // sum0 += (a20-a27) * k02 in conv_im2col_sgemm_sse()
309 _sum1 = _mm256_fmadd_ps(_vb2, _va1, _sum1); // sum1 += (a20-a27) * k12 in conv_im2col_sgemm_sse()
310 _sum2 = _mm256_fmadd_ps(_vb2, _va2, _sum2); // sum2 += (a20-a27) * k22 in conv_im2col_sgemm_sse()
535 __m256 _vb2 = _mm256_broadcast_ss(vb + 2); in conv_im2col_sgemm_sse() local
659 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
835 __m128 _vb2 = _mm_set1_ps(vb[2]); in conv_im2col_sgemm_sse() local
844 _sum2 = _mm_fmadd_ps(_va2, _vb2, _sum2); // sum2 += (k02-k32) * a20 in conv_im2col_sgemm_sse()
929 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
1403 __m128 _vb2 = _mm_set1_ps(vb[2]); in conv_im2col_sgemm_sse() local
[all …]
|
| /dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/x86/ |
| H A D | convolution_sgemm.h | 266 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
308 _sum0 = _mm256_fmadd_ps(_vb2, _va0, _sum0); // sum0 += (a20-a27) * k02 in conv_im2col_sgemm_sse()
309 _sum1 = _mm256_fmadd_ps(_vb2, _va1, _sum1); // sum1 += (a20-a27) * k12 in conv_im2col_sgemm_sse()
310 _sum2 = _mm256_fmadd_ps(_vb2, _va2, _sum2); // sum2 += (a20-a27) * k22 in conv_im2col_sgemm_sse()
535 __m256 _vb2 = _mm256_broadcast_ss(vb + 2); in conv_im2col_sgemm_sse() local
659 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
835 __m128 _vb2 = _mm_set1_ps(vb[2]); in conv_im2col_sgemm_sse() local
844 _sum2 = _mm_fmadd_ps(_va2, _vb2, _sum2); // sum2 += (k02-k32) * a20 in conv_im2col_sgemm_sse()
929 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
1403 __m128 _vb2 = _mm_set1_ps(vb[2]); in conv_im2col_sgemm_sse() local
[all …]
|
| /dports/misc/ncnn/ncnn-20211208/src/layer/x86/ |
| H A D | convolution_sgemm.h | 266 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
308 _sum0 = _mm256_comp_fmadd_ps(_vb2, _va0, _sum0); // sum0 += (a20-a27) * k02 in conv_im2col_sgemm_sse()
309 _sum1 = _mm256_comp_fmadd_ps(_vb2, _va1, _sum1); // sum1 += (a20-a27) * k12 in conv_im2col_sgemm_sse()
310 _sum2 = _mm256_comp_fmadd_ps(_vb2, _va2, _sum2); // sum2 += (a20-a27) * k22 in conv_im2col_sgemm_sse()
535 __m256 _vb2 = _mm256_broadcast_ss(vb + 2); in conv_im2col_sgemm_sse() local
659 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
835 __m128 _vb2 = _mm_set1_ps(vb[2]); in conv_im2col_sgemm_sse() local
844 _sum2 = _mm_comp_fmadd_ps(_va2, _vb2, _sum2); // sum2 += (k02-k32) * a20 in conv_im2col_sgemm_sse()
929 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
1403 __m128 _vb2 = _mm_set1_ps(vb[2]); 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 | 266 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
308 _sum0 = _mm256_fmadd_ps(_vb2, _va0, _sum0); // sum0 += (a20-a27) * k02 in conv_im2col_sgemm_sse()
309 _sum1 = _mm256_fmadd_ps(_vb2, _va1, _sum1); // sum1 += (a20-a27) * k12 in conv_im2col_sgemm_sse()
310 _sum2 = _mm256_fmadd_ps(_vb2, _va2, _sum2); // sum2 += (a20-a27) * k22 in conv_im2col_sgemm_sse()
535 __m256 _vb2 = _mm256_broadcast_ss(vb + 2); in conv_im2col_sgemm_sse() local
659 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
835 __m128 _vb2 = _mm_set1_ps(vb[2]); in conv_im2col_sgemm_sse() local
844 _sum2 = _mm_fmadd_ps(_va2, _vb2, _sum2); // sum2 += (k02-k32) * a20 in conv_im2col_sgemm_sse()
929 __m256 _vb2 = _mm256_loadu_ps(vb + 16); in conv_im2col_sgemm_sse() local
1403 __m128 _vb2 = _mm_set1_ps(vb[2]); in conv_im2col_sgemm_sse() local
[all …]
|