/dports/graphics/vapoursynth-waifu2x-ncnn-vulkan/vapoursynth-waifu2x-ncnn-vulkan-r4/deps/ncnn/src/layer/arm/ |
H A D | convolutiondepthwise_5x5_pack4.h | 234 float32x4_t _k34 = vld1q_f32(k0 + 16); in convdw5x5s1_pack4_neon() local 262 _sum00 = vmlaq_f32(_sum00, _k34, _r34); in convdw5x5s1_pack4_neon() 267 _sum01 = vmlaq_f32(_sum01, _k34, _r35); in convdw5x5s1_pack4_neon() 272 _sum02 = vmlaq_f32(_sum02, _k34, _r36); in convdw5x5s1_pack4_neon() 277 _sum03 = vmlaq_f32(_sum03, _k34, _r37); in convdw5x5s1_pack4_neon() 690 _sum0 = vmlaq_f32(_sum0, _k34, _r34); in convdw5x5s1_pack4_neon() 709 _sum1 = vmlaq_f32(_sum1, _k34, _r44); in convdw5x5s1_pack4_neon() 895 _sum0 = vmlaq_f32(_sum0, _k34, _r34); in convdw5x5s1_pack4_neon() 900 _sum1 = vmlaq_f32(_sum1, _k34, _r35); in convdw5x5s1_pack4_neon() 905 _sum2 = vmlaq_f32(_sum2, _k34, _r36); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1380 float16x8_t _k34 = vld1q_f16(k0 + 32); in convdw5x5s2_pack8_fp16sa_neon() local 1387 _sum0 = vfmaq_f16(_sum0, _k34, _r34); 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 | 234 float32x4_t _k34 = vld1q_f32(k0 + 16); in convdw5x5s1_pack4_neon() local 262 _sum00 = vmlaq_f32(_sum00, _k34, _r34); in convdw5x5s1_pack4_neon() 267 _sum01 = vmlaq_f32(_sum01, _k34, _r35); in convdw5x5s1_pack4_neon() 272 _sum02 = vmlaq_f32(_sum02, _k34, _r36); in convdw5x5s1_pack4_neon() 277 _sum03 = vmlaq_f32(_sum03, _k34, _r37); in convdw5x5s1_pack4_neon() 690 _sum0 = vmlaq_f32(_sum0, _k34, _r34); in convdw5x5s1_pack4_neon() 709 _sum1 = vmlaq_f32(_sum1, _k34, _r44); in convdw5x5s1_pack4_neon() 895 _sum0 = vmlaq_f32(_sum0, _k34, _r34); in convdw5x5s1_pack4_neon() 900 _sum1 = vmlaq_f32(_sum1, _k34, _r35); in convdw5x5s1_pack4_neon() 905 _sum2 = vmlaq_f32(_sum2, _k34, _r36); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1380 float16x8_t _k34 = vld1q_f16(k0 + 32); in convdw5x5s2_pack8_fp16sa_neon() local 1387 _sum0 = vfmaq_f16(_sum0, _k34, _r34); in convdw5x5s2_pack8_fp16sa_neon()
|
/dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/arm/ |
H A D | convolutiondepthwise_5x5_pack4.h | 234 float32x4_t _k34 = vld1q_f32(k0 + 16); in convdw5x5s1_pack4_neon() local 262 _sum00 = vmlaq_f32(_sum00, _k34, _r34); in convdw5x5s1_pack4_neon() 267 _sum01 = vmlaq_f32(_sum01, _k34, _r35); in convdw5x5s1_pack4_neon() 272 _sum02 = vmlaq_f32(_sum02, _k34, _r36); in convdw5x5s1_pack4_neon() 277 _sum03 = vmlaq_f32(_sum03, _k34, _r37); in convdw5x5s1_pack4_neon() 690 _sum0 = vmlaq_f32(_sum0, _k34, _r34); in convdw5x5s1_pack4_neon() 709 _sum1 = vmlaq_f32(_sum1, _k34, _r44); in convdw5x5s1_pack4_neon() 895 _sum0 = vmlaq_f32(_sum0, _k34, _r34); in convdw5x5s1_pack4_neon() 900 _sum1 = vmlaq_f32(_sum1, _k34, _r35); in convdw5x5s1_pack4_neon() 905 _sum2 = vmlaq_f32(_sum2, _k34, _r36); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1380 float16x8_t _k34 = vld1q_f16(k0 + 32); in convdw5x5s2_pack8_fp16sa_neon() local 1387 _sum0 = vfmaq_f16(_sum0, _k34, _r34); in convdw5x5s2_pack8_fp16sa_neon()
|
/dports/misc/ncnn/ncnn-20211208/src/layer/arm/ |
H A D | convolutiondepthwise_5x5_pack4.h | 234 float32x4_t _k34 = vld1q_f32(k0 + 16); in convdw5x5s1_pack4_neon() local 262 _sum00 = vmlaq_f32(_sum00, _k34, _r34); in convdw5x5s1_pack4_neon() 267 _sum01 = vmlaq_f32(_sum01, _k34, _r35); in convdw5x5s1_pack4_neon() 272 _sum02 = vmlaq_f32(_sum02, _k34, _r36); in convdw5x5s1_pack4_neon() 277 _sum03 = vmlaq_f32(_sum03, _k34, _r37); in convdw5x5s1_pack4_neon() 690 _sum0 = vmlaq_f32(_sum0, _k34, _r34); in convdw5x5s1_pack4_neon() 709 _sum1 = vmlaq_f32(_sum1, _k34, _r44); in convdw5x5s1_pack4_neon() 895 _sum0 = vmlaq_f32(_sum0, _k34, _r34); in convdw5x5s1_pack4_neon() 900 _sum1 = vmlaq_f32(_sum1, _k34, _r35); in convdw5x5s1_pack4_neon() 905 _sum2 = vmlaq_f32(_sum2, _k34, _r36); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1380 float16x8_t _k34 = vld1q_f16(k0 + 32); in convdw5x5s2_pack8_fp16sa_neon() local 1387 _sum0 = vfmaq_f16(_sum0, _k34, _r34); 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 | 234 float32x4_t _k34 = vld1q_f32(k0 + 16); in convdw5x5s1_pack4_neon() local 262 _sum00 = vmlaq_f32(_sum00, _k34, _r34); in convdw5x5s1_pack4_neon() 267 _sum01 = vmlaq_f32(_sum01, _k34, _r35); in convdw5x5s1_pack4_neon() 272 _sum02 = vmlaq_f32(_sum02, _k34, _r36); in convdw5x5s1_pack4_neon() 277 _sum03 = vmlaq_f32(_sum03, _k34, _r37); in convdw5x5s1_pack4_neon() 690 _sum0 = vmlaq_f32(_sum0, _k34, _r34); in convdw5x5s1_pack4_neon() 709 _sum1 = vmlaq_f32(_sum1, _k34, _r44); in convdw5x5s1_pack4_neon() 895 _sum0 = vmlaq_f32(_sum0, _k34, _r34); in convdw5x5s1_pack4_neon() 900 _sum1 = vmlaq_f32(_sum1, _k34, _r35); in convdw5x5s1_pack4_neon() 905 _sum2 = vmlaq_f32(_sum2, _k34, _r36); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1380 float16x8_t _k34 = vld1q_f16(k0 + 32); in convdw5x5s2_pack8_fp16sa_neon() local 1387 _sum0 = vfmaq_f16(_sum0, _k34, _r34); in convdw5x5s2_pack8_fp16sa_neon()
|
/dports/misc/ncnn/ncnn-20211208/src/layer/mips/ |
H A D | convolutiondepthwise_5x5_pack4.h | 150 v4f32 _k34 = (v4f32)__msa_ld_w(k0 + 4 * 4, 0); in convdw5x5s1_pack4_msa() local 157 _sum0 = __msa_fmadd_w(_sum0, _k34, _r34); in convdw5x5s1_pack4_msa() 169 _sum1 = __msa_fmadd_w(_sum1, _k34, _r44); in convdw5x5s1_pack4_msa() 302 v4f32 _k34 = (v4f32)__msa_ld_w(k0 + 4 * 4, 0); in convdw5x5s1_pack4_msa() local 309 _sum0 = __msa_fmadd_w(_sum0, _k34, _r34); in convdw5x5s1_pack4_msa() 465 v4f32 _k34 = (v4f32)__msa_ld_w(k0 + 4 * 4, 0); in convdw5x5s2_pack4_msa() local 472 _sum0 = __msa_fmadd_w(_sum0, _k34, _r34); in convdw5x5s2_pack4_msa()
|
H A D | convolution_7x7_pack1to4.h | 239 v4f32 _k34 = (v4f32)__msa_ld_w(kptr + 4 * 4, 0); in conv7x7s2_pack1to4_msa() local 279 _sum0 = __msa_fmadd_w(_sum0, _r34, _k34); in conv7x7s2_pack1to4_msa() 280 _sum1 = __msa_fmadd_w(_sum1, _r36, _k34); in conv7x7s2_pack1to4_msa() 281 _sum2 = __msa_fmadd_w(_sum2, _r38, _k34); in conv7x7s2_pack1to4_msa() 282 _sum3 = __msa_fmadd_w(_sum3, _r3a, _k34); in conv7x7s2_pack1to4_msa() 549 v4f32 _k34 = (v4f32)__msa_ld_w(kptr + 4 * 4, 0); in conv7x7s2_pack1to4_msa() local 562 _sum0 = __msa_fmadd_w(_sum0, (v4f32)__msa_splati_w(_r3n, 0), _k34); 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 | 118 __m256 _k34 = _mm256_loadu_ps(k0 + 32); in convdw5x5s1_pack8_avx() local 125 _sum0 = _mm256_fmadd_ps(_k34, _r34, _sum0); in convdw5x5s1_pack8_avx() 272 __m256 _k34 = _mm256_loadu_ps(k0 + 32); in convdw5x5s2_pack8_avx() local 279 _sum0 = _mm256_fmadd_ps(_k34, _r34, _sum0); in convdw5x5s2_pack8_avx()
|
H A D | convolution_3x3_pack8.h | 192 __m256 _k34 = _mm256_loadu_ps(kptr + 32); in conv3x3s1_pack8_avx() local 203 _sum00 = _mm256_fmadd_ps(_r104, _k34, _sum00); in conv3x3s1_pack8_avx() 221 _sum10 = _mm256_fmadd_ps(_r114, _k34, _sum10); in conv3x3s1_pack8_avx() 541 __m256 _k34 = _mm256_loadu_ps(kptr + 32); in conv3x3s1_pack8_avx() local 552 _sum0 = _mm256_fmadd_ps(_r104, _k34, _sum0); 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 | 118 __m256 _k34 = _mm256_loadu_ps(k0 + 32); in convdw5x5s1_pack8_avx() local 125 _sum0 = _mm256_fmadd_ps(_k34, _r34, _sum0); in convdw5x5s1_pack8_avx() 272 __m256 _k34 = _mm256_loadu_ps(k0 + 32); in convdw5x5s2_pack8_avx() local 279 _sum0 = _mm256_fmadd_ps(_k34, _r34, _sum0); in convdw5x5s2_pack8_avx()
|
H A D | convolution_3x3_pack8.h | 192 __m256 _k34 = _mm256_loadu_ps(kptr + 32); in conv3x3s1_pack8_avx() local 203 _sum00 = _mm256_fmadd_ps(_r104, _k34, _sum00); in conv3x3s1_pack8_avx() 221 _sum10 = _mm256_fmadd_ps(_r114, _k34, _sum10); in conv3x3s1_pack8_avx() 541 __m256 _k34 = _mm256_loadu_ps(kptr + 32); in conv3x3s1_pack8_avx() local 552 _sum0 = _mm256_fmadd_ps(_r104, _k34, _sum0); in conv3x3s1_pack8_avx()
|
/dports/misc/ncnn/ncnn-20211208/src/layer/x86/ |
H A D | convolutiondepthwise_5x5_pack8.h | 118 __m256 _k34 = _mm256_loadu_ps(k0 + 32); in convdw5x5s1_pack8_avx() local 125 _sum0 = _mm256_comp_fmadd_ps(_k34, _r34, _sum0); in convdw5x5s1_pack8_avx() 272 __m256 _k34 = _mm256_loadu_ps(k0 + 32); in convdw5x5s2_pack8_avx() local 279 _sum0 = _mm256_comp_fmadd_ps(_k34, _r34, _sum0); in convdw5x5s2_pack8_avx()
|
H A D | convolution_3x3_pack8.h | 192 __m256 _k34 = _mm256_loadu_ps(kptr + 32); in conv3x3s1_pack8_avx() local 203 _sum00 = _mm256_comp_fmadd_ps(_r104, _k34, _sum00); in conv3x3s1_pack8_avx() 221 _sum10 = _mm256_comp_fmadd_ps(_r114, _k34, _sum10); in conv3x3s1_pack8_avx() 541 __m256 _k34 = _mm256_loadu_ps(kptr + 32); in conv3x3s1_pack8_avx() local 552 _sum0 = _mm256_comp_fmadd_ps(_r104, _k34, _sum0); in conv3x3s1_pack8_avx()
|
/dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/x86/ |
H A D | convolutiondepthwise_5x5_pack8.h | 118 __m256 _k34 = _mm256_loadu_ps(k0 + 32); in convdw5x5s1_pack8_avx() local 125 _sum0 = _mm256_fmadd_ps(_k34, _r34, _sum0); in convdw5x5s1_pack8_avx() 272 __m256 _k34 = _mm256_loadu_ps(k0 + 32); in convdw5x5s2_pack8_avx() local 279 _sum0 = _mm256_fmadd_ps(_k34, _r34, _sum0); in convdw5x5s2_pack8_avx()
|
H A D | convolution_3x3_pack8.h | 192 __m256 _k34 = _mm256_loadu_ps(kptr + 32); in conv3x3s1_pack8_avx() local 203 _sum00 = _mm256_fmadd_ps(_r104, _k34, _sum00); in conv3x3s1_pack8_avx() 221 _sum10 = _mm256_fmadd_ps(_r114, _k34, _sum10); in conv3x3s1_pack8_avx() 541 __m256 _k34 = _mm256_loadu_ps(kptr + 32); in conv3x3s1_pack8_avx() local 552 _sum0 = _mm256_fmadd_ps(_r104, _k34, _sum0); 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 | 118 __m256 _k34 = _mm256_loadu_ps(k0 + 32); in convdw5x5s1_pack8_avx() local 125 _sum0 = _mm256_fmadd_ps(_k34, _r34, _sum0); in convdw5x5s1_pack8_avx() 272 __m256 _k34 = _mm256_loadu_ps(k0 + 32); in convdw5x5s2_pack8_avx() local 279 _sum0 = _mm256_fmadd_ps(_k34, _r34, _sum0); in convdw5x5s2_pack8_avx()
|
/dports/misc/ncnn/ncnn-20211208/src/layer/riscv/ |
H A D | convolutiondepthwise_5x5_packn.h | 144 vfloat32m1_t _k34 = vle32_v_f32m1(k0 + packn * 4, vl); in convdw5x5s1_packn_rvv() local 151 _sum0 = vfmacc_vv_f32m1(_sum0, _k34, _r34, vl); in convdw5x5s1_packn_rvv() 163 _sum1 = vfmacc_vv_f32m1(_sum1, _k34, _r44, vl); in convdw5x5s1_packn_rvv() 288 vfloat32m1_t _k34 = vle32_v_f32m1(k0 + packn * 4, vl); in convdw5x5s1_packn_rvv() local 295 _sum0 = vfmacc_vv_f32m1(_sum0, _k34, _r34, vl); in convdw5x5s1_packn_rvv() 446 vfloat32m1_t _k34 = vle32_v_f32m1(k0 + packn * 4, vl); in convdw5x5s2_packn_rvv() local 453 _sum0 = vfmacc_vv_f32m1(_sum0, _k34, _r34, vl); in convdw5x5s2_packn_rvv()
|
H A D | convolutiondepthwise_5x5_packn_fp16s.h | 144 vfloat16m1_t _k34 = vle16_v_f16m1(k0 + packn * 4, vl); in convdw5x5s1_packn_fp16sa_rvv() local 151 _sum0 = vfmacc_vv_f16m1(_sum0, _k34, _r34, vl); in convdw5x5s1_packn_fp16sa_rvv() 163 _sum1 = vfmacc_vv_f16m1(_sum1, _k34, _r44, vl); in convdw5x5s1_packn_fp16sa_rvv() 288 vfloat16m1_t _k34 = vle16_v_f16m1(k0 + packn * 4, vl); in convdw5x5s1_packn_fp16sa_rvv() local 295 _sum0 = vfmacc_vv_f16m1(_sum0, _k34, _r34, vl); in convdw5x5s1_packn_fp16sa_rvv() 446 vfloat16m1_t _k34 = vle16_v_f16m1(k0 + packn * 4, vl); in convdw5x5s2_packn_fp16sa_rvv() local 453 _sum0 = vfmacc_vv_f16m1(_sum0, _k34, _r34, vl); in convdw5x5s2_packn_fp16sa_rvv()
|
H A D | convolution_7x7_pack1ton.h | 314 _sum0 = vfmacc_vf_f32m1(_sum0, r3[4], _k34, vl); in conv7x7s2_pack1ton_rvv() 315 _sum1 = vfmacc_vf_f32m1(_sum1, r3[6], _k34, vl); in conv7x7s2_pack1ton_rvv() 316 _sum2 = vfmacc_vf_f32m1(_sum2, r3[8], _k34, vl); in conv7x7s2_pack1ton_rvv() 317 _sum3 = vfmacc_vf_f32m1(_sum3, r3[10], _k34, vl); in conv7x7s2_pack1ton_rvv() 318 _sum4 = vfmacc_vf_f32m1(_sum4, r3[12], _k34, vl); in conv7x7s2_pack1ton_rvv() 319 _sum5 = vfmacc_vf_f32m1(_sum5, r3[14], _k34, vl); in conv7x7s2_pack1ton_rvv() 320 _sum6 = vfmacc_vf_f32m1(_sum6, r3[16], _k34, vl); in conv7x7s2_pack1ton_rvv() 321 _sum7 = vfmacc_vf_f32m1(_sum7, r3[18], _k34, vl); in conv7x7s2_pack1ton_rvv() 709 _sum0 = vfmacc_vf_f32m1(_sum0, r3[4], _k34, vl); in conv7x7s2_pack1ton_rvv() 710 _sum1 = vfmacc_vf_f32m1(_sum1, r3[6], _k34, vl); in conv7x7s2_pack1ton_rvv() [all …]
|
H A D | convolution_7x7_pack1ton_fp16s.h | 314 _sum0 = vfmacc_vf_f16m1(_sum0, r3[4], _k34, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 315 _sum1 = vfmacc_vf_f16m1(_sum1, r3[6], _k34, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 316 _sum2 = vfmacc_vf_f16m1(_sum2, r3[8], _k34, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 317 _sum3 = vfmacc_vf_f16m1(_sum3, r3[10], _k34, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 318 _sum4 = vfmacc_vf_f16m1(_sum4, r3[12], _k34, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 319 _sum5 = vfmacc_vf_f16m1(_sum5, r3[14], _k34, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 320 _sum6 = vfmacc_vf_f16m1(_sum6, r3[16], _k34, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 321 _sum7 = vfmacc_vf_f16m1(_sum7, r3[18], _k34, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 709 _sum0 = vfmacc_vf_f16m1(_sum0, r3[4], _k34, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 710 _sum1 = vfmacc_vf_f16m1(_sum1, r3[6], _k34, vl); in conv7x7s2_pack1ton_fp16sa_rvv() [all …]
|