| /dports/graphics/vapoursynth-waifu2x-ncnn-vulkan/vapoursynth-waifu2x-ncnn-vulkan-r4/deps/ncnn/src/layer/arm/ |
| H A D | convolution_3x3_pack8to1_fp16s.h | 295 … float16x8_t _r0tm7 = vfmaq_n_f16(vsubq_f16(_tmp07, _tmp01), vsubq_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack8to1_fp16sa_neon() local
343 vst1q_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack8to1_fp16sa_neon()
|
| H A D | convolution_3x3_pack4to1_bf16s.h | 179 … float32x4_t _r0tm7 = vmlaq_n_f32(vsubq_f32(_tmp07, _tmp01), vsubq_f32(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4to1_bf16s_neon() local
227 vst1q_f32(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4to1_bf16s_neon()
|
| H A D | convolution_3x3_pack8to4_fp16s.h | 301 … float16x8_t _r0tm7 = vfmaq_n_f16(vsubq_f16(_tmp07, _tmp01), vsubq_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack8to4_fp16sa_neon() local
349 vst1q_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack8to4_fp16sa_neon()
|
| H A D | convolution_3x3_pack4_fp16s.h | 401 … float16x4_t _r0tm7 = vfma_n_f16(vsub_f16(_tmp07, _tmp01), vsub_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4_fp16sa_neon() local
449 vst1_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4_fp16sa_neon()
|
| H A D | convolution_3x3_pack4to1.h | 436 … float32x4_t _r0tm7 = vmlaq_n_f32(vsubq_f32(_tmp07, _tmp01), vsubq_f32(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4to1_neon() local
484 vst1q_f32(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4to1_neon()
|
| /dports/graphics/waifu2x-ncnn-vulkan/waifu2x-ncnn-vulkan-20210521/src/ncnn/src/layer/arm/ |
| H A D | convolution_3x3_pack8to1_fp16s.h | 295 … float16x8_t _r0tm7 = vfmaq_n_f16(vsubq_f16(_tmp07, _tmp01), vsubq_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack8to1_fp16sa_neon() local
343 vst1q_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack8to1_fp16sa_neon()
|
| H A D | convolution_3x3_pack4to1_bf16s.h | 179 … float32x4_t _r0tm7 = vmlaq_n_f32(vsubq_f32(_tmp07, _tmp01), vsubq_f32(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4to1_bf16s_neon() local
227 vst1q_f32(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4to1_bf16s_neon()
|
| H A D | convolution_3x3_pack8to4_fp16s.h | 301 … float16x8_t _r0tm7 = vfmaq_n_f16(vsubq_f16(_tmp07, _tmp01), vsubq_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack8to4_fp16sa_neon() local
349 vst1q_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack8to4_fp16sa_neon()
|
| H A D | convolution_3x3_pack4_fp16s.h | 401 … float16x4_t _r0tm7 = vfma_n_f16(vsub_f16(_tmp07, _tmp01), vsub_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4_fp16sa_neon() local
449 vst1_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4_fp16sa_neon()
|
| H A D | convolution_3x3_pack4to1.h | 436 … float32x4_t _r0tm7 = vmlaq_n_f32(vsubq_f32(_tmp07, _tmp01), vsubq_f32(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4to1_neon() local
484 vst1q_f32(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4to1_neon()
|
| /dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/arm/ |
| H A D | convolution_3x3_pack8to1_fp16s.h | 295 … float16x8_t _r0tm7 = vfmaq_n_f16(vsubq_f16(_tmp07, _tmp01), vsubq_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack8to1_fp16sa_neon() local
343 vst1q_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack8to1_fp16sa_neon()
|
| H A D | convolution_3x3_pack4to1_bf16s.h | 179 … float32x4_t _r0tm7 = vmlaq_n_f32(vsubq_f32(_tmp07, _tmp01), vsubq_f32(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4to1_bf16s_neon() local
227 vst1q_f32(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4to1_bf16s_neon()
|
| H A D | convolution_3x3_pack8to4_fp16s.h | 301 … float16x8_t _r0tm7 = vfmaq_n_f16(vsubq_f16(_tmp07, _tmp01), vsubq_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack8to4_fp16sa_neon() local
349 vst1q_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack8to4_fp16sa_neon()
|
| H A D | convolution_3x3_pack4_fp16s.h | 401 … float16x4_t _r0tm7 = vfma_n_f16(vsub_f16(_tmp07, _tmp01), vsub_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4_fp16sa_neon() local
449 vst1_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4_fp16sa_neon()
|
| H A D | convolution_3x3_pack4to1.h | 436 … float32x4_t _r0tm7 = vmlaq_n_f32(vsubq_f32(_tmp07, _tmp01), vsubq_f32(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4to1_neon() local
484 vst1q_f32(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4to1_neon()
|
| /dports/misc/ncnn/ncnn-20211208/src/layer/arm/ |
| H A D | convolution_3x3_pack8to1_fp16s.h | 295 … float16x8_t _r0tm7 = vfmaq_n_f16(vsubq_f16(_tmp07, _tmp01), vsubq_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack8to1_fp16sa_neon() local
343 vst1q_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack8to1_fp16sa_neon()
|
| H A D | convolution_3x3_pack4to1_bf16s.h | 179 … float32x4_t _r0tm7 = vmlaq_n_f32(vsubq_f32(_tmp07, _tmp01), vsubq_f32(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4to1_bf16s_neon() local
227 vst1q_f32(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4to1_bf16s_neon()
|
| H A D | convolution_3x3_pack8to4_fp16s.h | 301 … float16x8_t _r0tm7 = vfmaq_n_f16(vsubq_f16(_tmp07, _tmp01), vsubq_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack8to4_fp16sa_neon() local
349 vst1q_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack8to4_fp16sa_neon()
|
| H A D | convolution_3x3_pack4_fp16s.h | 401 … float16x4_t _r0tm7 = vfma_n_f16(vsub_f16(_tmp07, _tmp01), vsub_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4_fp16sa_neon() local
449 vst1_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4_fp16sa_neon()
|
| H A D | convolution_3x3_pack4to1.h | 436 … float32x4_t _r0tm7 = vmlaq_n_f32(vsubq_f32(_tmp07, _tmp01), vsubq_f32(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4to1_neon() local
484 vst1q_f32(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4to1_neon()
|
| /dports/graphics/realsr-ncnn-vulkan/realsr-ncnn-vulkan-20210210/src/ncnn/src/layer/arm/ |
| H A D | convolution_3x3_pack8to1_fp16s.h | 295 … float16x8_t _r0tm7 = vfmaq_n_f16(vsubq_f16(_tmp07, _tmp01), vsubq_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack8to1_fp16sa_neon() local
343 vst1q_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack8to1_fp16sa_neon()
|
| H A D | convolution_3x3_pack4to1_bf16s.h | 179 … float32x4_t _r0tm7 = vmlaq_n_f32(vsubq_f32(_tmp07, _tmp01), vsubq_f32(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4to1_bf16s_neon() local
227 vst1q_f32(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4to1_bf16s_neon()
|
| H A D | convolution_3x3_pack8to4_fp16s.h | 301 … float16x8_t _r0tm7 = vfmaq_n_f16(vsubq_f16(_tmp07, _tmp01), vsubq_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack8to4_fp16sa_neon() local
349 vst1q_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack8to4_fp16sa_neon()
|
| H A D | convolution_3x3_pack4_fp16s.h | 401 … float16x4_t _r0tm7 = vfma_n_f16(vsub_f16(_tmp07, _tmp01), vsub_f16(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4_fp16sa_neon() local
449 vst1_f16(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4_fp16sa_neon()
|
| H A D | convolution_3x3_pack4to1.h | 436 … float32x4_t _r0tm7 = vmlaq_n_f32(vsubq_f32(_tmp07, _tmp01), vsubq_f32(_tmp03, _tmp05), 5.25f); in conv3x3s1_winograd64_pack4to1_neon() local
484 vst1q_f32(r0_tm_7, _r0tm7); in conv3x3s1_winograd64_pack4to1_neon()
|