Searched refs:r0_tm3_4 (Results 1 – 5 of 5) sorted by relevance
/dports/graphics/waifu2x-ncnn-vulkan/waifu2x-ncnn-vulkan-20210521/src/ncnn/src/layer/arm/ |
H A D | convolution_3x3.h | 2055 float* r0_tm3_4 = img0_tm.row(i * w_tm / 8 + j + tiles * 7); in conv3x3s1_winograd64_neon4() local 2100 r0_tm3_4[3] = vgetq_lane_f32(_r0_tm_4_3, 3); in conv3x3s1_winograd64_neon4() 2141 r0_tm3_4[0] = vgetq_lane_f32(_r0_tm_4_0, 3); in conv3x3s1_winograd64_neon4() 2154 r0_tm3_4[1] = vgetq_lane_f32(_r0_tm_4_1, 3); in conv3x3s1_winograd64_neon4() 2159 r0_tm3_4[2] = vgetq_lane_f32(_r0_tm_4_2, 3); in conv3x3s1_winograd64_neon4() 2173 r0_tm3_4 += img0_tm.w * tiles * 2 * 4; in conv3x3s1_winograd64_neon4() 2386 float* r0_tm3_4 = img0_tm.row(i * w_tm / 8 + j + tiles * 7); in conv3x3s1_winograd64_neon4() local 2610 "=r"(r0_tm3_4), // %7 in conv3x3s1_winograd64_neon4() 2622 "7"(r0_tm3_4), in conv3x3s1_winograd64_neon4() 5379 "=r"(r0_tm3_4), // %7 in conv3x3s1_winograd64_neon5() [all …]
|
/dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/arm/ |
H A D | convolution_3x3.h | 2055 float* r0_tm3_4 = img0_tm.row(i * w_tm / 8 + j + tiles * 7); in conv3x3s1_winograd64_neon4() local 2100 r0_tm3_4[3] = vgetq_lane_f32(_r0_tm_4_3, 3); in conv3x3s1_winograd64_neon4() 2141 r0_tm3_4[0] = vgetq_lane_f32(_r0_tm_4_0, 3); in conv3x3s1_winograd64_neon4() 2154 r0_tm3_4[1] = vgetq_lane_f32(_r0_tm_4_1, 3); in conv3x3s1_winograd64_neon4() 2159 r0_tm3_4[2] = vgetq_lane_f32(_r0_tm_4_2, 3); in conv3x3s1_winograd64_neon4() 2173 r0_tm3_4 += img0_tm.w * tiles * 2 * 4; in conv3x3s1_winograd64_neon4() 2386 float* r0_tm3_4 = img0_tm.row(i * w_tm / 8 + j + tiles * 7); in conv3x3s1_winograd64_neon4() local 2610 "=r"(r0_tm3_4), // %7 in conv3x3s1_winograd64_neon4() 2622 "7"(r0_tm3_4), in conv3x3s1_winograd64_neon4() 5379 "=r"(r0_tm3_4), // %7 in conv3x3s1_winograd64_neon5() [all …]
|
/dports/misc/ncnn/ncnn-20211208/src/layer/arm/ |
H A D | convolution_3x3.h | 2055 float* r0_tm3_4 = img0_tm.row(i * w_tm / 8 + j + tiles * 7); in conv3x3s1_winograd64_neon4() local 2100 r0_tm3_4[3] = vgetq_lane_f32(_r0_tm_4_3, 3); in conv3x3s1_winograd64_neon4() 2141 r0_tm3_4[0] = vgetq_lane_f32(_r0_tm_4_0, 3); in conv3x3s1_winograd64_neon4() 2154 r0_tm3_4[1] = vgetq_lane_f32(_r0_tm_4_1, 3); in conv3x3s1_winograd64_neon4() 2159 r0_tm3_4[2] = vgetq_lane_f32(_r0_tm_4_2, 3); in conv3x3s1_winograd64_neon4() 2173 r0_tm3_4 += img0_tm.w * tiles * 2 * 4; in conv3x3s1_winograd64_neon4() 2386 float* r0_tm3_4 = img0_tm.row(i * w_tm / 8 + j + tiles * 7); in conv3x3s1_winograd64_neon4() local 2610 "=r"(r0_tm3_4), // %7 in conv3x3s1_winograd64_neon4() 2622 "7"(r0_tm3_4), in conv3x3s1_winograd64_neon4() 5379 "=r"(r0_tm3_4), // %7 in conv3x3s1_winograd64_neon5() [all …]
|
/dports/graphics/vapoursynth-waifu2x-ncnn-vulkan/vapoursynth-waifu2x-ncnn-vulkan-r4/deps/ncnn/src/layer/arm/ |
H A D | convolution_3x3.h | 2055 float* r0_tm3_4 = img0_tm.row(i * w_tm / 8 + j + tiles * 7); in conv3x3s1_winograd64_neon4() local 2100 r0_tm3_4[3] = vgetq_lane_f32(_r0_tm_4_3, 3); in conv3x3s1_winograd64_neon4() 2141 r0_tm3_4[0] = vgetq_lane_f32(_r0_tm_4_0, 3); in conv3x3s1_winograd64_neon4() 2154 r0_tm3_4[1] = vgetq_lane_f32(_r0_tm_4_1, 3); in conv3x3s1_winograd64_neon4() 2159 r0_tm3_4[2] = vgetq_lane_f32(_r0_tm_4_2, 3); in conv3x3s1_winograd64_neon4() 2173 r0_tm3_4 += img0_tm.w * tiles * 2 * 4; in conv3x3s1_winograd64_neon4() 2386 float* r0_tm3_4 = img0_tm.row(i * w_tm / 8 + j + tiles * 7); in conv3x3s1_winograd64_neon4() local 2610 "=r"(r0_tm3_4), // %7 in conv3x3s1_winograd64_neon4() 2622 "7"(r0_tm3_4), in conv3x3s1_winograd64_neon4() 5379 "=r"(r0_tm3_4), // %7 in conv3x3s1_winograd64_neon5() [all …]
|
/dports/graphics/realsr-ncnn-vulkan/realsr-ncnn-vulkan-20210210/src/ncnn/src/layer/arm/ |
H A D | convolution_3x3.h | 2055 float* r0_tm3_4 = img0_tm.row(i * w_tm / 8 + j + tiles * 7); in conv3x3s1_winograd64_neon4() local 2100 r0_tm3_4[3] = vgetq_lane_f32(_r0_tm_4_3, 3); in conv3x3s1_winograd64_neon4() 2141 r0_tm3_4[0] = vgetq_lane_f32(_r0_tm_4_0, 3); in conv3x3s1_winograd64_neon4() 2154 r0_tm3_4[1] = vgetq_lane_f32(_r0_tm_4_1, 3); in conv3x3s1_winograd64_neon4() 2159 r0_tm3_4[2] = vgetq_lane_f32(_r0_tm_4_2, 3); in conv3x3s1_winograd64_neon4() 2173 r0_tm3_4 += img0_tm.w * tiles * 2 * 4; in conv3x3s1_winograd64_neon4() 2386 float* r0_tm3_4 = img0_tm.row(i * w_tm / 8 + j + tiles * 7); in conv3x3s1_winograd64_neon4() local 2610 "=r"(r0_tm3_4), // %7 in conv3x3s1_winograd64_neon4() 2622 "7"(r0_tm3_4), in conv3x3s1_winograd64_neon4() 5379 "=r"(r0_tm3_4), // %7 in conv3x3s1_winograd64_neon5() [all …]
|