/dports/graphics/vapoursynth-waifu2x-ncnn-vulkan/vapoursynth-waifu2x-ncnn-vulkan-r4/deps/ncnn/src/layer/arm/ |
H A D | convolutiondepthwise_5x5_pack4.h | 291 float32x4_t _k43 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local 319 _sum00 = vmlaq_f32(_sum00, _k43, _r43); in convdw5x5s1_pack4_neon() 535 float32x4_t _k43 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local 714 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() 726 _sum1 = vmlaq_f32(_sum1, _k43, _r53); in convdw5x5s1_pack4_neon() 931 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() 936 _sum1 = vmlaq_f32(_sum1, _k43, _r44); in convdw5x5s1_pack4_neon() 941 _sum2 = vmlaq_f32(_sum2, _k43, _r45); in convdw5x5s1_pack4_neon() 946 _sum3 = vmlaq_f32(_sum3, _k43, _r46); in convdw5x5s1_pack4_neon() 1083 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1398 float16x8_t _k43 = vld1q_f16(k0 + 24); in convdw5x5s2_pack8_fp16sa_neon() local 1405 _sum0 = vfmaq_f16(_sum0, _k43, _r43); 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 | 291 float32x4_t _k43 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local 319 _sum00 = vmlaq_f32(_sum00, _k43, _r43); in convdw5x5s1_pack4_neon() 535 float32x4_t _k43 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local 714 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() 726 _sum1 = vmlaq_f32(_sum1, _k43, _r53); in convdw5x5s1_pack4_neon() 931 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() 936 _sum1 = vmlaq_f32(_sum1, _k43, _r44); in convdw5x5s1_pack4_neon() 941 _sum2 = vmlaq_f32(_sum2, _k43, _r45); in convdw5x5s1_pack4_neon() 946 _sum3 = vmlaq_f32(_sum3, _k43, _r46); in convdw5x5s1_pack4_neon() 1083 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1398 float16x8_t _k43 = vld1q_f16(k0 + 24); in convdw5x5s2_pack8_fp16sa_neon() local 1405 _sum0 = vfmaq_f16(_sum0, _k43, _r43); in convdw5x5s2_pack8_fp16sa_neon()
|
/dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/arm/ |
H A D | convolutiondepthwise_5x5_pack4.h | 291 float32x4_t _k43 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local 319 _sum00 = vmlaq_f32(_sum00, _k43, _r43); in convdw5x5s1_pack4_neon() 535 float32x4_t _k43 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local 714 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() 726 _sum1 = vmlaq_f32(_sum1, _k43, _r53); in convdw5x5s1_pack4_neon() 931 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() 936 _sum1 = vmlaq_f32(_sum1, _k43, _r44); in convdw5x5s1_pack4_neon() 941 _sum2 = vmlaq_f32(_sum2, _k43, _r45); in convdw5x5s1_pack4_neon() 946 _sum3 = vmlaq_f32(_sum3, _k43, _r46); in convdw5x5s1_pack4_neon() 1083 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1398 float16x8_t _k43 = vld1q_f16(k0 + 24); in convdw5x5s2_pack8_fp16sa_neon() local 1405 _sum0 = vfmaq_f16(_sum0, _k43, _r43); in convdw5x5s2_pack8_fp16sa_neon()
|
/dports/misc/ncnn/ncnn-20211208/src/layer/arm/ |
H A D | convolutiondepthwise_5x5_pack4.h | 291 float32x4_t _k43 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local 319 _sum00 = vmlaq_f32(_sum00, _k43, _r43); in convdw5x5s1_pack4_neon() 535 float32x4_t _k43 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local 714 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() 726 _sum1 = vmlaq_f32(_sum1, _k43, _r53); in convdw5x5s1_pack4_neon() 931 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() 936 _sum1 = vmlaq_f32(_sum1, _k43, _r44); in convdw5x5s1_pack4_neon() 941 _sum2 = vmlaq_f32(_sum2, _k43, _r45); in convdw5x5s1_pack4_neon() 946 _sum3 = vmlaq_f32(_sum3, _k43, _r46); in convdw5x5s1_pack4_neon() 1083 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1398 float16x8_t _k43 = vld1q_f16(k0 + 24); in convdw5x5s2_pack8_fp16sa_neon() local 1405 _sum0 = vfmaq_f16(_sum0, _k43, _r43); 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 | 291 float32x4_t _k43 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local 319 _sum00 = vmlaq_f32(_sum00, _k43, _r43); in convdw5x5s1_pack4_neon() 535 float32x4_t _k43 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local 714 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() 726 _sum1 = vmlaq_f32(_sum1, _k43, _r53); in convdw5x5s1_pack4_neon() 931 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() 936 _sum1 = vmlaq_f32(_sum1, _k43, _r44); in convdw5x5s1_pack4_neon() 941 _sum2 = vmlaq_f32(_sum2, _k43, _r45); in convdw5x5s1_pack4_neon() 946 _sum3 = vmlaq_f32(_sum3, _k43, _r46); in convdw5x5s1_pack4_neon() 1083 _sum0 = vmlaq_f32(_sum0, _k43, _r43); in convdw5x5s1_pack4_neon() [all …]
|
H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1398 float16x8_t _k43 = vld1q_f16(k0 + 24); in convdw5x5s2_pack8_fp16sa_neon() local 1405 _sum0 = vfmaq_f16(_sum0, _k43, _r43); in convdw5x5s2_pack8_fp16sa_neon()
|
/dports/misc/ncnn/ncnn-20211208/src/layer/mips/ |
H A D | convolutiondepthwise_5x5_pack4.h | 174 v4f32 _k43 = (v4f32)__msa_ld_w(k0 + 4 * 3, 0); in convdw5x5s1_pack4_msa() local 181 _sum0 = __msa_fmadd_w(_sum0, _k43, _r43); in convdw5x5s1_pack4_msa() 193 _sum1 = __msa_fmadd_w(_sum1, _k43, _r53); in convdw5x5s1_pack4_msa() 320 v4f32 _k43 = (v4f32)__msa_ld_w(k0 + 4 * 3, 0); in convdw5x5s1_pack4_msa() local 327 _sum0 = __msa_fmadd_w(_sum0, _k43, _r43); in convdw5x5s1_pack4_msa() 483 v4f32 _k43 = (v4f32)__msa_ld_w(k0 + 4 * 3, 0); in convdw5x5s2_pack4_msa() local 490 _sum0 = __msa_fmadd_w(_sum0, _k43, _r43); in convdw5x5s2_pack4_msa()
|
H A D | convolution_7x7_pack1to4.h | 295 v4f32 _k43 = (v4f32)__msa_ld_w(kptr + 4 * 3, 0); in conv7x7s2_pack1to4_msa() local 332 _sum0 = __msa_fmadd_w(_sum0, _r43, _k43); in conv7x7s2_pack1to4_msa() 333 _sum1 = __msa_fmadd_w(_sum1, _r45, _k43); in conv7x7s2_pack1to4_msa() 334 _sum2 = __msa_fmadd_w(_sum2, _r47, _k43); in conv7x7s2_pack1to4_msa() 335 _sum3 = __msa_fmadd_w(_sum3, _r49, _k43); in conv7x7s2_pack1to4_msa() 569 v4f32 _k43 = (v4f32)__msa_ld_w(kptr + 4 * 3, 0); in conv7x7s2_pack1to4_msa() local 582 _sum0 = __msa_fmadd_w(_sum0, (v4f32)__msa_splati_w(_r4, 3), _k43); 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 | 136 __m256 _k43 = _mm256_loadu_ps(k0 + 24); in convdw5x5s1_pack8_avx() local 143 _sum0 = _mm256_fmadd_ps(_k43, _r43, _sum0); in convdw5x5s1_pack8_avx() 290 __m256 _k43 = _mm256_loadu_ps(k0 + 24); in convdw5x5s2_pack8_avx() local 297 _sum0 = _mm256_fmadd_ps(_k43, _r43, _sum0); in convdw5x5s2_pack8_avx()
|
H A D | convolution_3x3_pack8.h | 229 __m256 _k43 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local 240 _sum01 = _mm256_fmadd_ps(_r113, _k43, _sum01); in conv3x3s1_pack8_avx() 258 _sum11 = _mm256_fmadd_ps(_r123, _k43, _sum11); in conv3x3s1_pack8_avx() 569 __m256 _k43 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local 580 _sum1 = _mm256_fmadd_ps(_r113, _k43, _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 | 136 __m256 _k43 = _mm256_loadu_ps(k0 + 24); in convdw5x5s1_pack8_avx() local 143 _sum0 = _mm256_fmadd_ps(_k43, _r43, _sum0); in convdw5x5s1_pack8_avx() 290 __m256 _k43 = _mm256_loadu_ps(k0 + 24); in convdw5x5s2_pack8_avx() local 297 _sum0 = _mm256_fmadd_ps(_k43, _r43, _sum0); in convdw5x5s2_pack8_avx()
|
H A D | convolution_3x3_pack8.h | 229 __m256 _k43 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local 240 _sum01 = _mm256_fmadd_ps(_r113, _k43, _sum01); in conv3x3s1_pack8_avx() 258 _sum11 = _mm256_fmadd_ps(_r123, _k43, _sum11); in conv3x3s1_pack8_avx() 569 __m256 _k43 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local 580 _sum1 = _mm256_fmadd_ps(_r113, _k43, _sum1); in conv3x3s1_pack8_avx()
|
/dports/misc/ncnn/ncnn-20211208/src/layer/x86/ |
H A D | convolutiondepthwise_5x5_pack8.h | 136 __m256 _k43 = _mm256_loadu_ps(k0 + 24); in convdw5x5s1_pack8_avx() local 143 _sum0 = _mm256_comp_fmadd_ps(_k43, _r43, _sum0); in convdw5x5s1_pack8_avx() 290 __m256 _k43 = _mm256_loadu_ps(k0 + 24); in convdw5x5s2_pack8_avx() local 297 _sum0 = _mm256_comp_fmadd_ps(_k43, _r43, _sum0); in convdw5x5s2_pack8_avx()
|
H A D | convolution_3x3_pack8.h | 229 __m256 _k43 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local 240 _sum01 = _mm256_comp_fmadd_ps(_r113, _k43, _sum01); in conv3x3s1_pack8_avx() 258 _sum11 = _mm256_comp_fmadd_ps(_r123, _k43, _sum11); in conv3x3s1_pack8_avx() 569 __m256 _k43 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local 580 _sum1 = _mm256_comp_fmadd_ps(_r113, _k43, _sum1); in conv3x3s1_pack8_avx()
|
/dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/x86/ |
H A D | convolutiondepthwise_5x5_pack8.h | 136 __m256 _k43 = _mm256_loadu_ps(k0 + 24); in convdw5x5s1_pack8_avx() local 143 _sum0 = _mm256_fmadd_ps(_k43, _r43, _sum0); in convdw5x5s1_pack8_avx() 290 __m256 _k43 = _mm256_loadu_ps(k0 + 24); in convdw5x5s2_pack8_avx() local 297 _sum0 = _mm256_fmadd_ps(_k43, _r43, _sum0); in convdw5x5s2_pack8_avx()
|
H A D | convolution_3x3_pack8.h | 229 __m256 _k43 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local 240 _sum01 = _mm256_fmadd_ps(_r113, _k43, _sum01); in conv3x3s1_pack8_avx() 258 _sum11 = _mm256_fmadd_ps(_r123, _k43, _sum11); in conv3x3s1_pack8_avx() 569 __m256 _k43 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local 580 _sum1 = _mm256_fmadd_ps(_r113, _k43, _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 | 136 __m256 _k43 = _mm256_loadu_ps(k0 + 24); in convdw5x5s1_pack8_avx() local 143 _sum0 = _mm256_fmadd_ps(_k43, _r43, _sum0); in convdw5x5s1_pack8_avx() 290 __m256 _k43 = _mm256_loadu_ps(k0 + 24); in convdw5x5s2_pack8_avx() local 297 _sum0 = _mm256_fmadd_ps(_k43, _r43, _sum0); in convdw5x5s2_pack8_avx()
|
/dports/misc/ncnn/ncnn-20211208/src/layer/riscv/ |
H A D | convolutiondepthwise_5x5_packn.h | 168 vfloat32m1_t _k43 = vle32_v_f32m1(k0 + packn * 3, vl); in convdw5x5s1_packn_rvv() local 175 _sum0 = vfmacc_vv_f32m1(_sum0, _k43, _r43, vl); in convdw5x5s1_packn_rvv() 187 _sum1 = vfmacc_vv_f32m1(_sum1, _k43, _r53, vl); in convdw5x5s1_packn_rvv() 306 vfloat32m1_t _k43 = vle32_v_f32m1(k0 + packn * 3, vl); in convdw5x5s1_packn_rvv() local 313 _sum0 = vfmacc_vv_f32m1(_sum0, _k43, _r43, vl); in convdw5x5s1_packn_rvv() 464 vfloat32m1_t _k43 = vle32_v_f32m1(k0 + packn * 3, vl); in convdw5x5s2_packn_rvv() local 471 _sum0 = vfmacc_vv_f32m1(_sum0, _k43, _r43, vl); in convdw5x5s2_packn_rvv()
|
H A D | convolutiondepthwise_5x5_packn_fp16s.h | 168 vfloat16m1_t _k43 = vle16_v_f16m1(k0 + packn * 3, vl); in convdw5x5s1_packn_fp16sa_rvv() local 175 _sum0 = vfmacc_vv_f16m1(_sum0, _k43, _r43, vl); in convdw5x5s1_packn_fp16sa_rvv() 187 _sum1 = vfmacc_vv_f16m1(_sum1, _k43, _r53, vl); in convdw5x5s1_packn_fp16sa_rvv() 306 vfloat16m1_t _k43 = vle16_v_f16m1(k0 + packn * 3, vl); in convdw5x5s1_packn_fp16sa_rvv() local 313 _sum0 = vfmacc_vv_f16m1(_sum0, _k43, _r43, vl); in convdw5x5s1_packn_fp16sa_rvv() 464 vfloat16m1_t _k43 = vle16_v_f16m1(k0 + packn * 3, vl); in convdw5x5s2_packn_fp16sa_rvv() local 471 _sum0 = vfmacc_vv_f16m1(_sum0, _k43, _r43, vl); in convdw5x5s2_packn_fp16sa_rvv()
|
H A D | convolution_7x7_pack1ton.h | 373 _sum0 = vfmacc_vf_f32m1(_sum0, r4[3], _k43, vl); in conv7x7s2_pack1ton_rvv() 374 _sum1 = vfmacc_vf_f32m1(_sum1, r4[5], _k43, vl); in conv7x7s2_pack1ton_rvv() 375 _sum2 = vfmacc_vf_f32m1(_sum2, r4[7], _k43, vl); in conv7x7s2_pack1ton_rvv() 376 _sum3 = vfmacc_vf_f32m1(_sum3, r4[9], _k43, vl); in conv7x7s2_pack1ton_rvv() 377 _sum4 = vfmacc_vf_f32m1(_sum4, r4[11], _k43, vl); in conv7x7s2_pack1ton_rvv() 378 _sum5 = vfmacc_vf_f32m1(_sum5, r4[13], _k43, vl); in conv7x7s2_pack1ton_rvv() 379 _sum6 = vfmacc_vf_f32m1(_sum6, r4[15], _k43, vl); in conv7x7s2_pack1ton_rvv() 380 _sum7 = vfmacc_vf_f32m1(_sum7, r4[17], _k43, vl); in conv7x7s2_pack1ton_rvv() 744 _sum0 = vfmacc_vf_f32m1(_sum0, r4[3], _k43, vl); in conv7x7s2_pack1ton_rvv() 745 _sum1 = vfmacc_vf_f32m1(_sum1, r4[5], _k43, vl); in conv7x7s2_pack1ton_rvv() [all …]
|
H A D | convolution_7x7_pack1ton_fp16s.h | 373 _sum0 = vfmacc_vf_f16m1(_sum0, r4[3], _k43, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 374 _sum1 = vfmacc_vf_f16m1(_sum1, r4[5], _k43, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 375 _sum2 = vfmacc_vf_f16m1(_sum2, r4[7], _k43, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 376 _sum3 = vfmacc_vf_f16m1(_sum3, r4[9], _k43, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 377 _sum4 = vfmacc_vf_f16m1(_sum4, r4[11], _k43, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 378 _sum5 = vfmacc_vf_f16m1(_sum5, r4[13], _k43, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 379 _sum6 = vfmacc_vf_f16m1(_sum6, r4[15], _k43, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 380 _sum7 = vfmacc_vf_f16m1(_sum7, r4[17], _k43, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 744 _sum0 = vfmacc_vf_f16m1(_sum0, r4[3], _k43, vl); in conv7x7s2_pack1ton_fp16sa_rvv() 745 _sum1 = vfmacc_vf_f16m1(_sum1, r4[5], _k43, vl); in conv7x7s2_pack1ton_fp16sa_rvv() [all …]
|