/dports/graphics/vapoursynth-waifu2x-ncnn-vulkan/vapoursynth-waifu2x-ncnn-vulkan-r4/deps/ncnn/src/layer/arm/ |
H A D | convolutiondepthwise_5x5_pack4.h | 231 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 259 _sum00 = vmlaq_f32(_sum00, _k31, _r31); in convdw5x5s1_pack4_neon() 264 _sum01 = vmlaq_f32(_sum01, _k31, _r32); in convdw5x5s1_pack4_neon() 497 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 674 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 687 _sum0 = vmlaq_f32(_sum0, _k31, _r31); in convdw5x5s1_pack4_neon() 706 _sum1 = vmlaq_f32(_sum1, _k31, _r41); in convdw5x5s1_pack4_neon() 885 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 892 _sum0 = vmlaq_f32(_sum0, _k31, _r31); in convdw5x5s1_pack4_neon() 897 _sum1 = vmlaq_f32(_sum1, _k31, _r32); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1377 float16x8_t _k31 = vld1q_f16(k0 + 8); in convdw5x5s2_pack8_fp16sa_neon() local 1384 _sum0 = vfmaq_f16(_sum0, _k31, _r31); in convdw5x5s2_pack8_fp16sa_neon()
|
/dports/graphics/waifu2x-ncnn-vulkan/waifu2x-ncnn-vulkan-20210521/src/ncnn/src/layer/arm/ |
H A D | convolutiondepthwise_5x5_pack4.h | 231 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 259 _sum00 = vmlaq_f32(_sum00, _k31, _r31); in convdw5x5s1_pack4_neon() 264 _sum01 = vmlaq_f32(_sum01, _k31, _r32); in convdw5x5s1_pack4_neon() 497 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 674 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 687 _sum0 = vmlaq_f32(_sum0, _k31, _r31); in convdw5x5s1_pack4_neon() 706 _sum1 = vmlaq_f32(_sum1, _k31, _r41); in convdw5x5s1_pack4_neon() 885 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 892 _sum0 = vmlaq_f32(_sum0, _k31, _r31); in convdw5x5s1_pack4_neon() 897 _sum1 = vmlaq_f32(_sum1, _k31, _r32); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1377 float16x8_t _k31 = vld1q_f16(k0 + 8); in convdw5x5s2_pack8_fp16sa_neon() local 1384 _sum0 = vfmaq_f16(_sum0, _k31, _r31); in convdw5x5s2_pack8_fp16sa_neon()
|
/dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/arm/ |
H A D | convolutiondepthwise_5x5_pack4.h | 231 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 259 _sum00 = vmlaq_f32(_sum00, _k31, _r31); in convdw5x5s1_pack4_neon() 264 _sum01 = vmlaq_f32(_sum01, _k31, _r32); in convdw5x5s1_pack4_neon() 497 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 674 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 687 _sum0 = vmlaq_f32(_sum0, _k31, _r31); in convdw5x5s1_pack4_neon() 706 _sum1 = vmlaq_f32(_sum1, _k31, _r41); in convdw5x5s1_pack4_neon() 885 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 892 _sum0 = vmlaq_f32(_sum0, _k31, _r31); in convdw5x5s1_pack4_neon() 897 _sum1 = vmlaq_f32(_sum1, _k31, _r32); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1377 float16x8_t _k31 = vld1q_f16(k0 + 8); in convdw5x5s2_pack8_fp16sa_neon() local 1384 _sum0 = vfmaq_f16(_sum0, _k31, _r31); in convdw5x5s2_pack8_fp16sa_neon()
|
/dports/misc/ncnn/ncnn-20211208/src/layer/arm/ |
H A D | convolutiondepthwise_5x5_pack4.h | 231 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 259 _sum00 = vmlaq_f32(_sum00, _k31, _r31); in convdw5x5s1_pack4_neon() 264 _sum01 = vmlaq_f32(_sum01, _k31, _r32); in convdw5x5s1_pack4_neon() 497 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 674 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 687 _sum0 = vmlaq_f32(_sum0, _k31, _r31); in convdw5x5s1_pack4_neon() 706 _sum1 = vmlaq_f32(_sum1, _k31, _r41); in convdw5x5s1_pack4_neon() 885 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 892 _sum0 = vmlaq_f32(_sum0, _k31, _r31); in convdw5x5s1_pack4_neon() 897 _sum1 = vmlaq_f32(_sum1, _k31, _r32); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1377 float16x8_t _k31 = vld1q_f16(k0 + 8); in convdw5x5s2_pack8_fp16sa_neon() local 1384 _sum0 = vfmaq_f16(_sum0, _k31, _r31); in convdw5x5s2_pack8_fp16sa_neon()
|
/dports/graphics/realsr-ncnn-vulkan/realsr-ncnn-vulkan-20210210/src/ncnn/src/layer/arm/ |
H A D | convolutiondepthwise_5x5_pack4.h | 231 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 259 _sum00 = vmlaq_f32(_sum00, _k31, _r31); in convdw5x5s1_pack4_neon() 264 _sum01 = vmlaq_f32(_sum01, _k31, _r32); in convdw5x5s1_pack4_neon() 497 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 674 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 687 _sum0 = vmlaq_f32(_sum0, _k31, _r31); in convdw5x5s1_pack4_neon() 706 _sum1 = vmlaq_f32(_sum1, _k31, _r41); in convdw5x5s1_pack4_neon() 885 float32x4_t _k31 = vld1q_f32(k0 + 4); in convdw5x5s1_pack4_neon() local 892 _sum0 = vmlaq_f32(_sum0, _k31, _r31); in convdw5x5s1_pack4_neon() 897 _sum1 = vmlaq_f32(_sum1, _k31, _r32); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1377 float16x8_t _k31 = vld1q_f16(k0 + 8); in convdw5x5s2_pack8_fp16sa_neon() local 1384 _sum0 = vfmaq_f16(_sum0, _k31, _r31); in convdw5x5s2_pack8_fp16sa_neon()
|
/dports/misc/ncnn/ncnn-20211208/src/layer/mips/ |
H A D | convolutiondepthwise_5x5_pack4.h | 147 v4f32 _k31 = (v4f32)__msa_ld_w(k0 + 4, 0); in convdw5x5s1_pack4_msa() local 154 _sum0 = __msa_fmadd_w(_sum0, _k31, _r31); in convdw5x5s1_pack4_msa() 166 _sum1 = __msa_fmadd_w(_sum1, _k31, _r41); in convdw5x5s1_pack4_msa() 299 v4f32 _k31 = (v4f32)__msa_ld_w(k0 + 4, 0); in convdw5x5s1_pack4_msa() local 306 _sum0 = __msa_fmadd_w(_sum0, _k31, _r31); in convdw5x5s1_pack4_msa() 462 v4f32 _k31 = (v4f32)__msa_ld_w(k0 + 4, 0); in convdw5x5s2_pack4_msa() local 469 _sum0 = __msa_fmadd_w(_sum0, _k31, _r31); in convdw5x5s2_pack4_msa()
|
H A D | convolution_7x7_pack1to4.h | 236 v4f32 _k31 = (v4f32)__msa_ld_w(kptr + 4, 0); in conv7x7s2_pack1to4_msa() local 267 _sum0 = __msa_fmadd_w(_sum0, _r31, _k31); in conv7x7s2_pack1to4_msa() 268 _sum1 = __msa_fmadd_w(_sum1, _r33, _k31); in conv7x7s2_pack1to4_msa() 269 _sum2 = __msa_fmadd_w(_sum2, _r35, _k31); in conv7x7s2_pack1to4_msa() 270 _sum3 = __msa_fmadd_w(_sum3, _r37, _k31); in conv7x7s2_pack1to4_msa() 546 v4f32 _k31 = (v4f32)__msa_ld_w(kptr + 4, 0); in conv7x7s2_pack1to4_msa() local 559 _sum0 = __msa_fmadd_w(_sum0, (v4f32)__msa_splati_w(_r3, 1), _k31); in conv7x7s2_pack1to4_msa()
|
/dports/graphics/vapoursynth-waifu2x-ncnn-vulkan/vapoursynth-waifu2x-ncnn-vulkan-r4/deps/ncnn/src/layer/x86/ |
H A D | convolutiondepthwise_5x5_pack8.h | 115 __m256 _k31 = _mm256_loadu_ps(k0 + 8); in convdw5x5s1_pack8_avx() local 122 _sum0 = _mm256_fmadd_ps(_k31, _r31, _sum0); in convdw5x5s1_pack8_avx() 269 __m256 _k31 = _mm256_loadu_ps(k0 + 8); in convdw5x5s2_pack8_avx() local 276 _sum0 = _mm256_fmadd_ps(_k31, _r31, _sum0); in convdw5x5s2_pack8_avx()
|
H A D | convolution_3x3_pack8.h | 189 __m256 _k31 = _mm256_loadu_ps(kptr + 8); in conv3x3s1_pack8_avx() local 200 _sum01 = _mm256_fmadd_ps(_r101, _k31, _sum01); in conv3x3s1_pack8_avx() 218 _sum11 = _mm256_fmadd_ps(_r111, _k31, _sum11); in conv3x3s1_pack8_avx() 538 __m256 _k31 = _mm256_loadu_ps(kptr + 8); in conv3x3s1_pack8_avx() local 549 _sum1 = _mm256_fmadd_ps(_r101, _k31, _sum1); in conv3x3s1_pack8_avx()
|
/dports/graphics/waifu2x-ncnn-vulkan/waifu2x-ncnn-vulkan-20210521/src/ncnn/src/layer/x86/ |
H A D | convolutiondepthwise_5x5_pack8.h | 115 __m256 _k31 = _mm256_loadu_ps(k0 + 8); in convdw5x5s1_pack8_avx() local 122 _sum0 = _mm256_fmadd_ps(_k31, _r31, _sum0); in convdw5x5s1_pack8_avx() 269 __m256 _k31 = _mm256_loadu_ps(k0 + 8); in convdw5x5s2_pack8_avx() local 276 _sum0 = _mm256_fmadd_ps(_k31, _r31, _sum0); in convdw5x5s2_pack8_avx()
|
H A D | convolution_3x3_pack8.h | 189 __m256 _k31 = _mm256_loadu_ps(kptr + 8); in conv3x3s1_pack8_avx() local 200 _sum01 = _mm256_fmadd_ps(_r101, _k31, _sum01); in conv3x3s1_pack8_avx() 218 _sum11 = _mm256_fmadd_ps(_r111, _k31, _sum11); in conv3x3s1_pack8_avx() 538 __m256 _k31 = _mm256_loadu_ps(kptr + 8); in conv3x3s1_pack8_avx() local 549 _sum1 = _mm256_fmadd_ps(_r101, _k31, _sum1); in conv3x3s1_pack8_avx()
|
/dports/misc/ncnn/ncnn-20211208/src/layer/x86/ |
H A D | convolutiondepthwise_5x5_pack8.h | 115 __m256 _k31 = _mm256_loadu_ps(k0 + 8); in convdw5x5s1_pack8_avx() local 122 _sum0 = _mm256_comp_fmadd_ps(_k31, _r31, _sum0); in convdw5x5s1_pack8_avx() 269 __m256 _k31 = _mm256_loadu_ps(k0 + 8); in convdw5x5s2_pack8_avx() local 276 _sum0 = _mm256_comp_fmadd_ps(_k31, _r31, _sum0); in convdw5x5s2_pack8_avx()
|
H A D | convolution_3x3_pack8.h | 189 __m256 _k31 = _mm256_loadu_ps(kptr + 8); in conv3x3s1_pack8_avx() local 200 _sum01 = _mm256_comp_fmadd_ps(_r101, _k31, _sum01); in conv3x3s1_pack8_avx() 218 _sum11 = _mm256_comp_fmadd_ps(_r111, _k31, _sum11); in conv3x3s1_pack8_avx() 538 __m256 _k31 = _mm256_loadu_ps(kptr + 8); in conv3x3s1_pack8_avx() local 549 _sum1 = _mm256_comp_fmadd_ps(_r101, _k31, _sum1); in conv3x3s1_pack8_avx()
|
/dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/x86/ |
H A D | convolutiondepthwise_5x5_pack8.h | 115 __m256 _k31 = _mm256_loadu_ps(k0 + 8); in convdw5x5s1_pack8_avx() local 122 _sum0 = _mm256_fmadd_ps(_k31, _r31, _sum0); in convdw5x5s1_pack8_avx() 269 __m256 _k31 = _mm256_loadu_ps(k0 + 8); in convdw5x5s2_pack8_avx() local 276 _sum0 = _mm256_fmadd_ps(_k31, _r31, _sum0); in convdw5x5s2_pack8_avx()
|
H A D | convolution_3x3_pack8.h | 189 __m256 _k31 = _mm256_loadu_ps(kptr + 8); in conv3x3s1_pack8_avx() local 200 _sum01 = _mm256_fmadd_ps(_r101, _k31, _sum01); in conv3x3s1_pack8_avx() 218 _sum11 = _mm256_fmadd_ps(_r111, _k31, _sum11); in conv3x3s1_pack8_avx() 538 __m256 _k31 = _mm256_loadu_ps(kptr + 8); in conv3x3s1_pack8_avx() local 549 _sum1 = _mm256_fmadd_ps(_r101, _k31, _sum1); in conv3x3s1_pack8_avx()
|
/dports/graphics/realsr-ncnn-vulkan/realsr-ncnn-vulkan-20210210/src/ncnn/src/layer/x86/ |
H A D | convolutiondepthwise_5x5_pack8.h | 115 __m256 _k31 = _mm256_loadu_ps(k0 + 8); in convdw5x5s1_pack8_avx() local 122 _sum0 = _mm256_fmadd_ps(_k31, _r31, _sum0); in convdw5x5s1_pack8_avx() 269 __m256 _k31 = _mm256_loadu_ps(k0 + 8); in convdw5x5s2_pack8_avx() local 276 _sum0 = _mm256_fmadd_ps(_k31, _r31, _sum0); in convdw5x5s2_pack8_avx()
|
/dports/misc/ncnn/ncnn-20211208/src/layer/riscv/ |
H A D | convolutiondepthwise_5x5_packn.h | 141 vfloat32m1_t _k31 = vle32_v_f32m1(k0 + packn, vl); in convdw5x5s1_packn_rvv() local 148 _sum0 = vfmacc_vv_f32m1(_sum0, _k31, _r31, vl); in convdw5x5s1_packn_rvv() 160 _sum1 = vfmacc_vv_f32m1(_sum1, _k31, _r41, vl); in convdw5x5s1_packn_rvv() 285 vfloat32m1_t _k31 = vle32_v_f32m1(k0 + packn, vl); in convdw5x5s1_packn_rvv() local 292 _sum0 = vfmacc_vv_f32m1(_sum0, _k31, _r31, vl); in convdw5x5s1_packn_rvv() 443 vfloat32m1_t _k31 = vle32_v_f32m1(k0 + packn, vl); in convdw5x5s2_packn_rvv() local 450 _sum0 = vfmacc_vv_f32m1(_sum0, _k31, _r31, vl); in convdw5x5s2_packn_rvv()
|
H A D | convolutiondepthwise_5x5_packn_fp16s.h | 141 vfloat16m1_t _k31 = vle16_v_f16m1(k0 + packn, vl); in convdw5x5s1_packn_fp16sa_rvv() local 148 _sum0 = vfmacc_vv_f16m1(_sum0, _k31, _r31, vl); in convdw5x5s1_packn_fp16sa_rvv() 160 _sum1 = vfmacc_vv_f16m1(_sum1, _k31, _r41, vl); in convdw5x5s1_packn_fp16sa_rvv() 285 vfloat16m1_t _k31 = vle16_v_f16m1(k0 + packn, vl); in convdw5x5s1_packn_fp16sa_rvv() local 292 _sum0 = vfmacc_vv_f16m1(_sum0, _k31, _r31, vl); in convdw5x5s1_packn_fp16sa_rvv() 443 vfloat16m1_t _k31 = vle16_v_f16m1(k0 + packn, vl); in convdw5x5s2_packn_fp16sa_rvv() local 450 _sum0 = vfmacc_vv_f16m1(_sum0, _k31, _r31, vl); in convdw5x5s2_packn_fp16sa_rvv()
|
H A D | convolution_7x7_pack1ton.h | 290 _sum0 = vfmacc_vf_f32m1(_sum0, r3[1], _k31, vl); in conv7x7s2_pack1ton_rvv() 291 _sum1 = vfmacc_vf_f32m1(_sum1, r3[3], _k31, vl); in conv7x7s2_pack1ton_rvv() 292 _sum2 = vfmacc_vf_f32m1(_sum2, r3[5], _k31, vl); in conv7x7s2_pack1ton_rvv() 293 _sum3 = vfmacc_vf_f32m1(_sum3, r3[7], _k31, vl); in conv7x7s2_pack1ton_rvv() 294 _sum4 = vfmacc_vf_f32m1(_sum4, r3[9], _k31, vl); in conv7x7s2_pack1ton_rvv() 295 _sum5 = vfmacc_vf_f32m1(_sum5, r3[11], _k31, vl); in conv7x7s2_pack1ton_rvv() 296 _sum6 = vfmacc_vf_f32m1(_sum6, r3[13], _k31, vl); in conv7x7s2_pack1ton_rvv() 297 _sum7 = vfmacc_vf_f32m1(_sum7, r3[15], _k31, vl); in conv7x7s2_pack1ton_rvv() 697 _sum0 = vfmacc_vf_f32m1(_sum0, r3[1], _k31, vl); in conv7x7s2_pack1ton_rvv() 698 _sum1 = vfmacc_vf_f32m1(_sum1, r3[3], _k31, vl); in conv7x7s2_pack1ton_rvv() [all …]
|
H A D | convolution_7x7_pack1ton_fp16s.h | 290 _sum0 = vfmacc_vf_f16m1(_sum0, r3[1], _k31, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 291 _sum1 = vfmacc_vf_f16m1(_sum1, r3[3], _k31, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 292 _sum2 = vfmacc_vf_f16m1(_sum2, r3[5], _k31, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 293 _sum3 = vfmacc_vf_f16m1(_sum3, r3[7], _k31, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 294 _sum4 = vfmacc_vf_f16m1(_sum4, r3[9], _k31, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 295 _sum5 = vfmacc_vf_f16m1(_sum5, r3[11], _k31, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 296 _sum6 = vfmacc_vf_f16m1(_sum6, r3[13], _k31, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 297 _sum7 = vfmacc_vf_f16m1(_sum7, r3[15], _k31, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 697 _sum0 = vfmacc_vf_f16m1(_sum0, r3[1], _k31, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 698 _sum1 = vfmacc_vf_f16m1(_sum1, r3[3], _k31, vl); in conv7x7s2_pack1ton_fp16sa_rvv() [all …]
|