| /dports/graphics/vapoursynth-waifu2x-ncnn-vulkan/vapoursynth-waifu2x-ncnn-vulkan-r4/deps/ncnn/src/layer/x86/ |
| H A D | convolution_sgemm.h | 265 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
288 _sum0 = _mm256_fmadd_ps(_vb1, _va0, _sum0); // sum0 += (a10-a17) * k01 in conv_im2col_sgemm_sse()
289 _sum1 = _mm256_fmadd_ps(_vb1, _va1, _sum1); // sum1 += (a10-a17) * k11 in conv_im2col_sgemm_sse()
290 _sum2 = _mm256_fmadd_ps(_vb1, _va2, _sum2); // sum2 += (a10-a17) * k21 in conv_im2col_sgemm_sse()
534 __m256 _vb1 = _mm256_broadcast_ss(vb + 1); in conv_im2col_sgemm_sse() local
658 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
834 __m128 _vb1 = _mm_set1_ps(vb[1]); in conv_im2col_sgemm_sse() local
843 _sum1 = _mm_fmadd_ps(_va1, _vb1, _sum1); // sum1 += (k01-k31) * a10 in conv_im2col_sgemm_sse()
928 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
1402 __m128 _vb1 = _mm_set1_ps(vb[1]); 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 | 265 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
288 _sum0 = _mm256_fmadd_ps(_vb1, _va0, _sum0); // sum0 += (a10-a17) * k01 in conv_im2col_sgemm_sse()
289 _sum1 = _mm256_fmadd_ps(_vb1, _va1, _sum1); // sum1 += (a10-a17) * k11 in conv_im2col_sgemm_sse()
290 _sum2 = _mm256_fmadd_ps(_vb1, _va2, _sum2); // sum2 += (a10-a17) * k21 in conv_im2col_sgemm_sse()
534 __m256 _vb1 = _mm256_broadcast_ss(vb + 1); in conv_im2col_sgemm_sse() local
658 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
834 __m128 _vb1 = _mm_set1_ps(vb[1]); in conv_im2col_sgemm_sse() local
843 _sum1 = _mm_fmadd_ps(_va1, _vb1, _sum1); // sum1 += (k01-k31) * a10 in conv_im2col_sgemm_sse()
928 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
1402 __m128 _vb1 = _mm_set1_ps(vb[1]); in conv_im2col_sgemm_sse() local
[all …]
|
| /dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/x86/ |
| H A D | convolution_sgemm.h | 265 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
288 _sum0 = _mm256_fmadd_ps(_vb1, _va0, _sum0); // sum0 += (a10-a17) * k01 in conv_im2col_sgemm_sse()
289 _sum1 = _mm256_fmadd_ps(_vb1, _va1, _sum1); // sum1 += (a10-a17) * k11 in conv_im2col_sgemm_sse()
290 _sum2 = _mm256_fmadd_ps(_vb1, _va2, _sum2); // sum2 += (a10-a17) * k21 in conv_im2col_sgemm_sse()
534 __m256 _vb1 = _mm256_broadcast_ss(vb + 1); in conv_im2col_sgemm_sse() local
658 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
834 __m128 _vb1 = _mm_set1_ps(vb[1]); in conv_im2col_sgemm_sse() local
843 _sum1 = _mm_fmadd_ps(_va1, _vb1, _sum1); // sum1 += (k01-k31) * a10 in conv_im2col_sgemm_sse()
928 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
1402 __m128 _vb1 = _mm_set1_ps(vb[1]); in conv_im2col_sgemm_sse() local
[all …]
|
| /dports/misc/ncnn/ncnn-20211208/src/layer/x86/ |
| H A D | convolution_sgemm.h | 265 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
288 _sum0 = _mm256_comp_fmadd_ps(_vb1, _va0, _sum0); // sum0 += (a10-a17) * k01 in conv_im2col_sgemm_sse()
289 _sum1 = _mm256_comp_fmadd_ps(_vb1, _va1, _sum1); // sum1 += (a10-a17) * k11 in conv_im2col_sgemm_sse()
290 _sum2 = _mm256_comp_fmadd_ps(_vb1, _va2, _sum2); // sum2 += (a10-a17) * k21 in conv_im2col_sgemm_sse()
534 __m256 _vb1 = _mm256_broadcast_ss(vb + 1); in conv_im2col_sgemm_sse() local
658 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
834 __m128 _vb1 = _mm_set1_ps(vb[1]); in conv_im2col_sgemm_sse() local
843 _sum1 = _mm_comp_fmadd_ps(_va1, _vb1, _sum1); // sum1 += (k01-k31) * a10 in conv_im2col_sgemm_sse()
928 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
1402 __m128 _vb1 = _mm_set1_ps(vb[1]); 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 | 265 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
288 _sum0 = _mm256_fmadd_ps(_vb1, _va0, _sum0); // sum0 += (a10-a17) * k01 in conv_im2col_sgemm_sse()
289 _sum1 = _mm256_fmadd_ps(_vb1, _va1, _sum1); // sum1 += (a10-a17) * k11 in conv_im2col_sgemm_sse()
290 _sum2 = _mm256_fmadd_ps(_vb1, _va2, _sum2); // sum2 += (a10-a17) * k21 in conv_im2col_sgemm_sse()
534 __m256 _vb1 = _mm256_broadcast_ss(vb + 1); in conv_im2col_sgemm_sse() local
658 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
834 __m128 _vb1 = _mm_set1_ps(vb[1]); in conv_im2col_sgemm_sse() local
843 _sum1 = _mm_fmadd_ps(_va1, _vb1, _sum1); // sum1 += (k01-k31) * a10 in conv_im2col_sgemm_sse()
928 __m256 _vb1 = _mm256_loadu_ps(vb + 8); in conv_im2col_sgemm_sse() local
1402 __m128 _vb1 = _mm_set1_ps(vb[1]); in conv_im2col_sgemm_sse() local
[all …]
|