| /dports/graphics/vapoursynth-waifu2x-ncnn-vulkan/vapoursynth-waifu2x-ncnn-vulkan-r4/deps/ncnn/src/layer/arm/ |
| H A D | convolutiondepthwise_5x5_pack4.h | 233 float32x4_t _k33 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local
261 _sum00 = vmlaq_f32(_sum00, _k33, _r33); in convdw5x5s1_pack4_neon()
266 _sum01 = vmlaq_f32(_sum01, _k33, _r34); in convdw5x5s1_pack4_neon()
271 _sum02 = vmlaq_f32(_sum02, _k33, _r35); in convdw5x5s1_pack4_neon()
276 _sum03 = vmlaq_f32(_sum03, _k33, _r36); in convdw5x5s1_pack4_neon()
689 _sum0 = vmlaq_f32(_sum0, _k33, _r33); in convdw5x5s1_pack4_neon()
708 _sum1 = vmlaq_f32(_sum1, _k33, _r43); in convdw5x5s1_pack4_neon()
894 _sum0 = vmlaq_f32(_sum0, _k33, _r33); in convdw5x5s1_pack4_neon()
899 _sum1 = vmlaq_f32(_sum1, _k33, _r34); in convdw5x5s1_pack4_neon()
904 _sum2 = vmlaq_f32(_sum2, _k33, _r35); in convdw5x5s1_pack4_neon()
[all …]
|
| H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1379 float16x8_t _k33 = vld1q_f16(k0 + 24); in convdw5x5s2_pack8_fp16sa_neon() local
1386 _sum0 = vfmaq_f16(_sum0, _k33, _r33); 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 | 233 float32x4_t _k33 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local
261 _sum00 = vmlaq_f32(_sum00, _k33, _r33); in convdw5x5s1_pack4_neon()
266 _sum01 = vmlaq_f32(_sum01, _k33, _r34); in convdw5x5s1_pack4_neon()
271 _sum02 = vmlaq_f32(_sum02, _k33, _r35); in convdw5x5s1_pack4_neon()
276 _sum03 = vmlaq_f32(_sum03, _k33, _r36); in convdw5x5s1_pack4_neon()
689 _sum0 = vmlaq_f32(_sum0, _k33, _r33); in convdw5x5s1_pack4_neon()
708 _sum1 = vmlaq_f32(_sum1, _k33, _r43); in convdw5x5s1_pack4_neon()
894 _sum0 = vmlaq_f32(_sum0, _k33, _r33); in convdw5x5s1_pack4_neon()
899 _sum1 = vmlaq_f32(_sum1, _k33, _r34); in convdw5x5s1_pack4_neon()
904 _sum2 = vmlaq_f32(_sum2, _k33, _r35); in convdw5x5s1_pack4_neon()
[all …]
|
| H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1379 float16x8_t _k33 = vld1q_f16(k0 + 24); in convdw5x5s2_pack8_fp16sa_neon() local
1386 _sum0 = vfmaq_f16(_sum0, _k33, _r33); in convdw5x5s2_pack8_fp16sa_neon()
|
| /dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/arm/ |
| H A D | convolutiondepthwise_5x5_pack4.h | 233 float32x4_t _k33 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local
261 _sum00 = vmlaq_f32(_sum00, _k33, _r33); in convdw5x5s1_pack4_neon()
266 _sum01 = vmlaq_f32(_sum01, _k33, _r34); in convdw5x5s1_pack4_neon()
271 _sum02 = vmlaq_f32(_sum02, _k33, _r35); in convdw5x5s1_pack4_neon()
276 _sum03 = vmlaq_f32(_sum03, _k33, _r36); in convdw5x5s1_pack4_neon()
689 _sum0 = vmlaq_f32(_sum0, _k33, _r33); in convdw5x5s1_pack4_neon()
708 _sum1 = vmlaq_f32(_sum1, _k33, _r43); in convdw5x5s1_pack4_neon()
894 _sum0 = vmlaq_f32(_sum0, _k33, _r33); in convdw5x5s1_pack4_neon()
899 _sum1 = vmlaq_f32(_sum1, _k33, _r34); in convdw5x5s1_pack4_neon()
904 _sum2 = vmlaq_f32(_sum2, _k33, _r35); in convdw5x5s1_pack4_neon()
[all …]
|
| H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1379 float16x8_t _k33 = vld1q_f16(k0 + 24); in convdw5x5s2_pack8_fp16sa_neon() local
1386 _sum0 = vfmaq_f16(_sum0, _k33, _r33); in convdw5x5s2_pack8_fp16sa_neon()
|
| /dports/misc/ncnn/ncnn-20211208/src/layer/arm/ |
| H A D | convolutiondepthwise_5x5_pack4.h | 233 float32x4_t _k33 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local
261 _sum00 = vmlaq_f32(_sum00, _k33, _r33); in convdw5x5s1_pack4_neon()
266 _sum01 = vmlaq_f32(_sum01, _k33, _r34); in convdw5x5s1_pack4_neon()
271 _sum02 = vmlaq_f32(_sum02, _k33, _r35); in convdw5x5s1_pack4_neon()
276 _sum03 = vmlaq_f32(_sum03, _k33, _r36); in convdw5x5s1_pack4_neon()
689 _sum0 = vmlaq_f32(_sum0, _k33, _r33); in convdw5x5s1_pack4_neon()
708 _sum1 = vmlaq_f32(_sum1, _k33, _r43); in convdw5x5s1_pack4_neon()
894 _sum0 = vmlaq_f32(_sum0, _k33, _r33); in convdw5x5s1_pack4_neon()
899 _sum1 = vmlaq_f32(_sum1, _k33, _r34); in convdw5x5s1_pack4_neon()
904 _sum2 = vmlaq_f32(_sum2, _k33, _r35); in convdw5x5s1_pack4_neon()
[all …]
|
| H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1379 float16x8_t _k33 = vld1q_f16(k0 + 24); in convdw5x5s2_pack8_fp16sa_neon() local
1386 _sum0 = vfmaq_f16(_sum0, _k33, _r33); 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 | 233 float32x4_t _k33 = vld1q_f32(k0 + 12); in convdw5x5s1_pack4_neon() local
261 _sum00 = vmlaq_f32(_sum00, _k33, _r33); in convdw5x5s1_pack4_neon()
266 _sum01 = vmlaq_f32(_sum01, _k33, _r34); in convdw5x5s1_pack4_neon()
271 _sum02 = vmlaq_f32(_sum02, _k33, _r35); in convdw5x5s1_pack4_neon()
276 _sum03 = vmlaq_f32(_sum03, _k33, _r36); in convdw5x5s1_pack4_neon()
689 _sum0 = vmlaq_f32(_sum0, _k33, _r33); in convdw5x5s1_pack4_neon()
708 _sum1 = vmlaq_f32(_sum1, _k33, _r43); in convdw5x5s1_pack4_neon()
894 _sum0 = vmlaq_f32(_sum0, _k33, _r33); in convdw5x5s1_pack4_neon()
899 _sum1 = vmlaq_f32(_sum1, _k33, _r34); in convdw5x5s1_pack4_neon()
904 _sum2 = vmlaq_f32(_sum2, _k33, _r35); in convdw5x5s1_pack4_neon()
[all …]
|
| H A D | convolutiondepthwise_5x5_pack8_fp16s.h | 1379 float16x8_t _k33 = vld1q_f16(k0 + 24); in convdw5x5s2_pack8_fp16sa_neon() local
1386 _sum0 = vfmaq_f16(_sum0, _k33, _r33); in convdw5x5s2_pack8_fp16sa_neon()
|
| /dports/misc/ncnn/ncnn-20211208/src/layer/mips/ |
| H A D | convolutiondepthwise_5x5_pack4.h | 149 v4f32 _k33 = (v4f32)__msa_ld_w(k0 + 4 * 3, 0); in convdw5x5s1_pack4_msa() local
156 _sum0 = __msa_fmadd_w(_sum0, _k33, _r33); in convdw5x5s1_pack4_msa()
168 _sum1 = __msa_fmadd_w(_sum1, _k33, _r43); in convdw5x5s1_pack4_msa()
301 v4f32 _k33 = (v4f32)__msa_ld_w(k0 + 4 * 3, 0); in convdw5x5s1_pack4_msa() local
308 _sum0 = __msa_fmadd_w(_sum0, _k33, _r33); in convdw5x5s1_pack4_msa()
464 v4f32 _k33 = (v4f32)__msa_ld_w(k0 + 4 * 3, 0); in convdw5x5s2_pack4_msa() local
471 _sum0 = __msa_fmadd_w(_sum0, _k33, _r33); in convdw5x5s2_pack4_msa()
|
| H A D | convolution_7x7_pack1to4.h | 238 v4f32 _k33 = (v4f32)__msa_ld_w(kptr + 4 * 3, 0); in conv7x7s2_pack1to4_msa() local
275 _sum0 = __msa_fmadd_w(_sum0, _r33, _k33); in conv7x7s2_pack1to4_msa()
276 _sum1 = __msa_fmadd_w(_sum1, _r35, _k33); in conv7x7s2_pack1to4_msa()
277 _sum2 = __msa_fmadd_w(_sum2, _r37, _k33); in conv7x7s2_pack1to4_msa()
278 _sum3 = __msa_fmadd_w(_sum3, _r39, _k33); in conv7x7s2_pack1to4_msa()
548 v4f32 _k33 = (v4f32)__msa_ld_w(kptr + 4 * 3, 0); in conv7x7s2_pack1to4_msa() local
561 _sum0 = __msa_fmadd_w(_sum0, (v4f32)__msa_splati_w(_r3, 3), _k33); 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 | 117 __m256 _k33 = _mm256_loadu_ps(k0 + 24); in convdw5x5s1_pack8_avx() local
124 _sum0 = _mm256_fmadd_ps(_k33, _r33, _sum0); in convdw5x5s1_pack8_avx()
271 __m256 _k33 = _mm256_loadu_ps(k0 + 24); in convdw5x5s2_pack8_avx() local
278 _sum0 = _mm256_fmadd_ps(_k33, _r33, _sum0); in convdw5x5s2_pack8_avx()
|
| H A D | convolution_3x3_pack8.h | 191 __m256 _k33 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local
202 _sum01 = _mm256_fmadd_ps(_r103, _k33, _sum01); in conv3x3s1_pack8_avx()
220 _sum11 = _mm256_fmadd_ps(_r113, _k33, _sum11); in conv3x3s1_pack8_avx()
540 __m256 _k33 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local
551 _sum1 = _mm256_fmadd_ps(_r103, _k33, _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 | 117 __m256 _k33 = _mm256_loadu_ps(k0 + 24); in convdw5x5s1_pack8_avx() local
124 _sum0 = _mm256_fmadd_ps(_k33, _r33, _sum0); in convdw5x5s1_pack8_avx()
271 __m256 _k33 = _mm256_loadu_ps(k0 + 24); in convdw5x5s2_pack8_avx() local
278 _sum0 = _mm256_fmadd_ps(_k33, _r33, _sum0); in convdw5x5s2_pack8_avx()
|
| H A D | convolution_3x3_pack8.h | 191 __m256 _k33 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local
202 _sum01 = _mm256_fmadd_ps(_r103, _k33, _sum01); in conv3x3s1_pack8_avx()
220 _sum11 = _mm256_fmadd_ps(_r113, _k33, _sum11); in conv3x3s1_pack8_avx()
540 __m256 _k33 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local
551 _sum1 = _mm256_fmadd_ps(_r103, _k33, _sum1); in conv3x3s1_pack8_avx()
|
| /dports/misc/ncnn/ncnn-20211208/src/layer/x86/ |
| H A D | convolutiondepthwise_5x5_pack8.h | 117 __m256 _k33 = _mm256_loadu_ps(k0 + 24); in convdw5x5s1_pack8_avx() local
124 _sum0 = _mm256_comp_fmadd_ps(_k33, _r33, _sum0); in convdw5x5s1_pack8_avx()
271 __m256 _k33 = _mm256_loadu_ps(k0 + 24); in convdw5x5s2_pack8_avx() local
278 _sum0 = _mm256_comp_fmadd_ps(_k33, _r33, _sum0); in convdw5x5s2_pack8_avx()
|
| H A D | convolution_3x3_pack8.h | 191 __m256 _k33 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local
202 _sum01 = _mm256_comp_fmadd_ps(_r103, _k33, _sum01); in conv3x3s1_pack8_avx()
220 _sum11 = _mm256_comp_fmadd_ps(_r113, _k33, _sum11); in conv3x3s1_pack8_avx()
540 __m256 _k33 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local
551 _sum1 = _mm256_comp_fmadd_ps(_r103, _k33, _sum1); in conv3x3s1_pack8_avx()
|
| /dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/x86/ |
| H A D | convolutiondepthwise_5x5_pack8.h | 117 __m256 _k33 = _mm256_loadu_ps(k0 + 24); in convdw5x5s1_pack8_avx() local
124 _sum0 = _mm256_fmadd_ps(_k33, _r33, _sum0); in convdw5x5s1_pack8_avx()
271 __m256 _k33 = _mm256_loadu_ps(k0 + 24); in convdw5x5s2_pack8_avx() local
278 _sum0 = _mm256_fmadd_ps(_k33, _r33, _sum0); in convdw5x5s2_pack8_avx()
|
| H A D | convolution_3x3_pack8.h | 191 __m256 _k33 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local
202 _sum01 = _mm256_fmadd_ps(_r103, _k33, _sum01); in conv3x3s1_pack8_avx()
220 _sum11 = _mm256_fmadd_ps(_r113, _k33, _sum11); in conv3x3s1_pack8_avx()
540 __m256 _k33 = _mm256_loadu_ps(kptr + 24); in conv3x3s1_pack8_avx() local
551 _sum1 = _mm256_fmadd_ps(_r103, _k33, _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 | 117 __m256 _k33 = _mm256_loadu_ps(k0 + 24); in convdw5x5s1_pack8_avx() local
124 _sum0 = _mm256_fmadd_ps(_k33, _r33, _sum0); in convdw5x5s1_pack8_avx()
271 __m256 _k33 = _mm256_loadu_ps(k0 + 24); in convdw5x5s2_pack8_avx() local
278 _sum0 = _mm256_fmadd_ps(_k33, _r33, _sum0); in convdw5x5s2_pack8_avx()
|
| /dports/misc/ncnn/ncnn-20211208/src/layer/riscv/ |
| H A D | convolutiondepthwise_5x5_packn.h | 143 vfloat32m1_t _k33 = vle32_v_f32m1(k0 + packn * 3, vl); in convdw5x5s1_packn_rvv() local
150 _sum0 = vfmacc_vv_f32m1(_sum0, _k33, _r33, vl); in convdw5x5s1_packn_rvv()
162 _sum1 = vfmacc_vv_f32m1(_sum1, _k33, _r43, vl); in convdw5x5s1_packn_rvv()
287 vfloat32m1_t _k33 = vle32_v_f32m1(k0 + packn * 3, vl); in convdw5x5s1_packn_rvv() local
294 _sum0 = vfmacc_vv_f32m1(_sum0, _k33, _r33, vl); in convdw5x5s1_packn_rvv()
445 vfloat32m1_t _k33 = vle32_v_f32m1(k0 + packn * 3, vl); in convdw5x5s2_packn_rvv() local
452 _sum0 = vfmacc_vv_f32m1(_sum0, _k33, _r33, vl); in convdw5x5s2_packn_rvv()
|
| H A D | convolutiondepthwise_5x5_packn_fp16s.h | 143 vfloat16m1_t _k33 = vle16_v_f16m1(k0 + packn * 3, vl); in convdw5x5s1_packn_fp16sa_rvv() local
150 _sum0 = vfmacc_vv_f16m1(_sum0, _k33, _r33, vl); in convdw5x5s1_packn_fp16sa_rvv()
162 _sum1 = vfmacc_vv_f16m1(_sum1, _k33, _r43, vl); in convdw5x5s1_packn_fp16sa_rvv()
287 vfloat16m1_t _k33 = vle16_v_f16m1(k0 + packn * 3, vl); in convdw5x5s1_packn_fp16sa_rvv() local
294 _sum0 = vfmacc_vv_f16m1(_sum0, _k33, _r33, vl); in convdw5x5s1_packn_fp16sa_rvv()
445 vfloat16m1_t _k33 = vle16_v_f16m1(k0 + packn * 3, vl); in convdw5x5s2_packn_fp16sa_rvv() local
452 _sum0 = vfmacc_vv_f16m1(_sum0, _k33, _r33, vl); in convdw5x5s2_packn_fp16sa_rvv()
|
| H A D | convolution_7x7_pack1ton.h | 306 _sum0 = vfmacc_vf_f32m1(_sum0, r3[3], _k33, vl); in conv7x7s2_pack1ton_rvv()
307 _sum1 = vfmacc_vf_f32m1(_sum1, r3[5], _k33, vl); in conv7x7s2_pack1ton_rvv()
308 _sum2 = vfmacc_vf_f32m1(_sum2, r3[7], _k33, vl); in conv7x7s2_pack1ton_rvv()
309 _sum3 = vfmacc_vf_f32m1(_sum3, r3[9], _k33, vl); in conv7x7s2_pack1ton_rvv()
310 _sum4 = vfmacc_vf_f32m1(_sum4, r3[11], _k33, vl); in conv7x7s2_pack1ton_rvv()
311 _sum5 = vfmacc_vf_f32m1(_sum5, r3[13], _k33, vl); in conv7x7s2_pack1ton_rvv()
312 _sum6 = vfmacc_vf_f32m1(_sum6, r3[15], _k33, vl); in conv7x7s2_pack1ton_rvv()
313 _sum7 = vfmacc_vf_f32m1(_sum7, r3[17], _k33, vl); in conv7x7s2_pack1ton_rvv()
705 _sum0 = vfmacc_vf_f32m1(_sum0, r3[3], _k33, vl); in conv7x7s2_pack1ton_rvv()
706 _sum1 = vfmacc_vf_f32m1(_sum1, r3[5], _k33, vl); in conv7x7s2_pack1ton_rvv()
[all …]
|
| H A D | convolution_7x7_pack1ton_fp16s.h | 306 _sum0 = vfmacc_vf_f16m1(_sum0, r3[3], _k33, vl); in conv7x7s2_pack1ton_fp16sa_rvv()
307 _sum1 = vfmacc_vf_f16m1(_sum1, r3[5], _k33, vl); in conv7x7s2_pack1ton_fp16sa_rvv()
308 _sum2 = vfmacc_vf_f16m1(_sum2, r3[7], _k33, vl); in conv7x7s2_pack1ton_fp16sa_rvv()
309 _sum3 = vfmacc_vf_f16m1(_sum3, r3[9], _k33, vl); in conv7x7s2_pack1ton_fp16sa_rvv()
310 _sum4 = vfmacc_vf_f16m1(_sum4, r3[11], _k33, vl); in conv7x7s2_pack1ton_fp16sa_rvv()
311 _sum5 = vfmacc_vf_f16m1(_sum5, r3[13], _k33, vl); in conv7x7s2_pack1ton_fp16sa_rvv()
312 _sum6 = vfmacc_vf_f16m1(_sum6, r3[15], _k33, vl); in conv7x7s2_pack1ton_fp16sa_rvv()
313 _sum7 = vfmacc_vf_f16m1(_sum7, r3[17], _k33, vl); in conv7x7s2_pack1ton_fp16sa_rvv()
705 _sum0 = vfmacc_vf_f16m1(_sum0, r3[3], _k33, vl); in conv7x7s2_pack1ton_fp16sa_rvv()
706 _sum1 = vfmacc_vf_f16m1(_sum1, r3[5], _k33, vl); in conv7x7s2_pack1ton_fp16sa_rvv()
[all …]
|