Home
last modified time | relevance | path

Searched refs:sum3_2 (Results 1 – 12 of 12) sorted by relevance

/dports/graphics/realsr-ncnn-vulkan/realsr-ncnn-vulkan-20210210/src/ncnn/src/layer/arm/
H A Dconvolution_1x1_int8.h444 int sum3_2 = 0; in conv1x1s1_sgemm_int8_neon() local
481 sum3_2 += tmpptr[4] * kptr[6]; in conv1x1s1_sgemm_int8_neon()
482 sum3_2 += tmpptr[5] * kptr[7]; in conv1x1s1_sgemm_int8_neon()
533 outptr3[2] = sum3_2; in conv1x1s1_sgemm_int8_neon()
1103 int sum3_2 = 0; in conv1x1s1_sgemm_int8_requant_neon() local
1784 int sum3_2 = 0; in conv1x1s1_sgemm_int8_neon() local
1862 outptr3[2] = sum3_2; in conv1x1s1_sgemm_int8_neon()
1986 int sum3_2 = 0; in conv1x1s1_sgemm_int8_neon() local
2032 outptr3[2] = sum3_2; in conv1x1s1_sgemm_int8_neon()
2819 int sum3_2 = 0; in conv1x1s1_sgemm_int8_requant_neon() local
[all …]
H A Dconvolution_1x1_bf16s.h1178 float sum3_2 = biasptr[3]; in conv1x1s1_sgemm_bf16s_neon() local
1216 sum3_2 += bfloat16_to_float32(tmpptr[2]) * bfloat16_to_float32(kptr[3]); in conv1x1s1_sgemm_bf16s_neon()
1256 outptr3[2] = float32_to_bfloat16(sum3_2); in conv1x1s1_sgemm_bf16s_neon()
1519 float sum3_2 = biasptr[3]; in conv1x1s1_sgemm_bf16s_neon() local
1541 sum3_2 += bfloat16_to_float32(tmpptr[2]) * bfloat16_to_float32(kptr[3]); in conv1x1s1_sgemm_bf16s_neon()
1565 outptr3[2] = float32_to_bfloat16(sum3_2); in conv1x1s1_sgemm_bf16s_neon()
H A Dconvolution_1x1.h1032 float sum3_2 = biasptr[3]; in conv1x1s1_sgemm_neon() local
1070 sum3_2 += tmpptr[2] * kptr[3]; in conv1x1s1_sgemm_neon()
1110 outptr3[2] = sum3_2; in conv1x1s1_sgemm_neon()
1339 float sum3_2 = biasptr[3]; in conv1x1s1_sgemm_neon() local
1361 sum3_2 += tmpptr[2] * kptr[3]; in conv1x1s1_sgemm_neon()
1385 outptr3[2] = sum3_2; in conv1x1s1_sgemm_neon()
H A Dconvolution_3x3.h6328 float sum3_2 = 0.f; in conv3x3s1_winograd64_neon5() local
6366 sum3_2 += bb2p0[2] * ktm0[3]; in conv3x3s1_winograd64_neon5()
6406 output3_tm[2] = sum3_2; in conv3x3s1_winograd64_neon5()
6624 float sum3_2 = 0.f; in conv3x3s1_winograd64_neon5() local
6646 sum3_2 += bb2p0[2] * ktm0[3]; in conv3x3s1_winograd64_neon5()
6670 output3_tm[2] = sum3_2; in conv3x3s1_winograd64_neon5()
/dports/graphics/waifu2x-ncnn-vulkan/waifu2x-ncnn-vulkan-20210521/src/ncnn/src/layer/arm/
H A Dconvolution_1x1_bf16s.h1178 float sum3_2 = biasptr[3]; in conv1x1s1_sgemm_bf16s_neon() local
1216 sum3_2 += bfloat16_to_float32(tmpptr[2]) * bfloat16_to_float32(kptr[3]); in conv1x1s1_sgemm_bf16s_neon()
1256 outptr3[2] = float32_to_bfloat16(sum3_2); in conv1x1s1_sgemm_bf16s_neon()
1519 float sum3_2 = biasptr[3]; in conv1x1s1_sgemm_bf16s_neon() local
1541 sum3_2 += bfloat16_to_float32(tmpptr[2]) * bfloat16_to_float32(kptr[3]); in conv1x1s1_sgemm_bf16s_neon()
1565 outptr3[2] = float32_to_bfloat16(sum3_2); in conv1x1s1_sgemm_bf16s_neon()
H A Dconvolution_3x3.h6328 float sum3_2 = 0.f; in conv3x3s1_winograd64_neon5() local
6366 sum3_2 += bb2p0[2] * ktm0[3]; in conv3x3s1_winograd64_neon5()
6406 output3_tm[2] = sum3_2; in conv3x3s1_winograd64_neon5()
6624 float sum3_2 = 0.f; in conv3x3s1_winograd64_neon5() local
6646 sum3_2 += bb2p0[2] * ktm0[3]; in conv3x3s1_winograd64_neon5()
6670 output3_tm[2] = sum3_2; in conv3x3s1_winograd64_neon5()
/dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/arm/
H A Dconvolution_1x1_bf16s.h1178 float sum3_2 = biasptr[3]; in conv1x1s1_sgemm_bf16s_neon() local
1216 sum3_2 += bfloat16_to_float32(tmpptr[2]) * bfloat16_to_float32(kptr[3]); in conv1x1s1_sgemm_bf16s_neon()
1256 outptr3[2] = float32_to_bfloat16(sum3_2); in conv1x1s1_sgemm_bf16s_neon()
1519 float sum3_2 = biasptr[3]; in conv1x1s1_sgemm_bf16s_neon() local
1541 sum3_2 += bfloat16_to_float32(tmpptr[2]) * bfloat16_to_float32(kptr[3]); in conv1x1s1_sgemm_bf16s_neon()
1565 outptr3[2] = float32_to_bfloat16(sum3_2); in conv1x1s1_sgemm_bf16s_neon()
H A Dconvolution_3x3.h6328 float sum3_2 = 0.f; in conv3x3s1_winograd64_neon5() local
6366 sum3_2 += bb2p0[2] * ktm0[3]; in conv3x3s1_winograd64_neon5()
6406 output3_tm[2] = sum3_2; in conv3x3s1_winograd64_neon5()
6624 float sum3_2 = 0.f; in conv3x3s1_winograd64_neon5() local
6646 sum3_2 += bb2p0[2] * ktm0[3]; in conv3x3s1_winograd64_neon5()
6670 output3_tm[2] = sum3_2; in conv3x3s1_winograd64_neon5()
/dports/misc/ncnn/ncnn-20211208/src/layer/arm/
H A Dconvolution_1x1_bf16s.h1178 float sum3_2 = biasptr[3]; in conv1x1s1_sgemm_bf16s_neon() local
1216 sum3_2 += bfloat16_to_float32(tmpptr[2]) * bfloat16_to_float32(kptr[3]); in conv1x1s1_sgemm_bf16s_neon()
1256 outptr3[2] = float32_to_bfloat16(sum3_2); in conv1x1s1_sgemm_bf16s_neon()
1519 float sum3_2 = biasptr[3]; in conv1x1s1_sgemm_bf16s_neon() local
1541 sum3_2 += bfloat16_to_float32(tmpptr[2]) * bfloat16_to_float32(kptr[3]); in conv1x1s1_sgemm_bf16s_neon()
1565 outptr3[2] = float32_to_bfloat16(sum3_2); in conv1x1s1_sgemm_bf16s_neon()
H A Dconvolution_3x3.h6328 float sum3_2 = 0.f; in conv3x3s1_winograd64_neon5() local
6366 sum3_2 += bb2p0[2] * ktm0[3]; in conv3x3s1_winograd64_neon5()
6406 output3_tm[2] = sum3_2; in conv3x3s1_winograd64_neon5()
6624 float sum3_2 = 0.f; in conv3x3s1_winograd64_neon5() local
6646 sum3_2 += bb2p0[2] * ktm0[3]; in conv3x3s1_winograd64_neon5()
6670 output3_tm[2] = sum3_2; in conv3x3s1_winograd64_neon5()
/dports/graphics/vapoursynth-waifu2x-ncnn-vulkan/vapoursynth-waifu2x-ncnn-vulkan-r4/deps/ncnn/src/layer/arm/
H A Dconvolution_1x1_bf16s.h1178 float sum3_2 = biasptr[3]; in conv1x1s1_sgemm_bf16s_neon() local
1216 sum3_2 += bfloat16_to_float32(tmpptr[2]) * bfloat16_to_float32(kptr[3]); in conv1x1s1_sgemm_bf16s_neon()
1256 outptr3[2] = float32_to_bfloat16(sum3_2); in conv1x1s1_sgemm_bf16s_neon()
1519 float sum3_2 = biasptr[3]; in conv1x1s1_sgemm_bf16s_neon() local
1541 sum3_2 += bfloat16_to_float32(tmpptr[2]) * bfloat16_to_float32(kptr[3]); in conv1x1s1_sgemm_bf16s_neon()
1565 outptr3[2] = float32_to_bfloat16(sum3_2); in conv1x1s1_sgemm_bf16s_neon()
H A Dconvolution_3x3.h6328 float sum3_2 = 0.f; in conv3x3s1_winograd64_neon5() local
6366 sum3_2 += bb2p0[2] * ktm0[3]; in conv3x3s1_winograd64_neon5()
6406 output3_tm[2] = sum3_2; in conv3x3s1_winograd64_neon5()
6624 float sum3_2 = 0.f; in conv3x3s1_winograd64_neon5() local
6646 sum3_2 += bb2p0[2] * ktm0[3]; in conv3x3s1_winograd64_neon5()
6670 output3_tm[2] = sum3_2; in conv3x3s1_winograd64_neon5()