109       load_src_0 = _mm256_loadu_ps(input_ptr);                         \
110       load_src_1 = _mm256_loadu_ps(input_ptr + 7);                     \
117       input_ptr += in_stride;                                          \
156     load_src[0] = _mm_loadu_ps(input_ptr);                               \
157     last_column_sum += input_ptr[4] * weight[0][4];                      \
158     input_ptr += in_stride;                                              \
159     load_src[1] = _mm_loadu_ps(input_ptr);                               \
160     last_column_sum += input_ptr[4] * weight[1][4];                      \
161     input_ptr += in_stride;                                              \
162     load_src[2] = _mm_loadu_ps(input_ptr);                               \
163     last_column_sum += input_ptr[4] * weight[2][4];                      \
164     input_ptr += in_stride;                                              \
165     load_src[3] = _mm_loadu_ps(input_ptr);                               \
166     last_column_sum += input_ptr[4] * weight[3][4];                      \
167     input_ptr += in_stride;                                              \
168     load_src[4] = _mm_loadu_ps(input_ptr);                               \
169     last_column_sum += input_ptr[4] * weight[4][4];                      \
186     const float *input_ptr, int in_stride, __m256 *weight, __m256 *out_accum,  in perform_convolve_for_8h_2x2_blocks()  argument
190   load_src[0] = _mm256_loadu_ps(input_ptr);  in perform_convolve_for_8h_2x2_blocks()
191   load_src[1] = _mm256_loadu_ps(input_ptr + 8);  in perform_convolve_for_8h_2x2_blocks()
192   load_src[2] = _mm256_loadu_ps(input_ptr + in_stride);  in perform_convolve_for_8h_2x2_blocks()
193   load_src[3] = _mm256_loadu_ps(input_ptr + in_stride + 8);  in perform_convolve_for_8h_2x2_blocks()
213     const float *input_ptr, int in_stride, __m256 *weight, __m256 *out_accum,  in perform_convolve_for_4hx2v_2x2_blocks()  argument
217   load_src[0] = _mm256_loadu_ps(input_ptr);  in perform_convolve_for_4hx2v_2x2_blocks()
218   load_src[1] = _mm256_loadu_ps(input_ptr + in_stride);  in perform_convolve_for_4hx2v_2x2_blocks()
219   load_src[2] = _mm256_loadu_ps(input_ptr + (in_stride * 2));  in perform_convolve_for_4hx2v_2x2_blocks()
220   load_src[3] = _mm256_loadu_ps(input_ptr + (in_stride * 3));  in perform_convolve_for_4hx2v_2x2_blocks()
306           const float *input_ptr = &input[k][h * in_stride + w];  in cnn_convolve_no_maxpool_padding_valid_5x5_avx2()  local
347           const float *input_ptr = &input[k][h * in_stride + w];  in cnn_convolve_no_maxpool_padding_valid_5x5_avx2()  local
400         const float *input_ptr = &input[k][h * in_stride];  in cnn_convolve_no_maxpool_padding_valid_layer1_avx2()  local
401         perform_convolve_for_8h_2x2_blocks(input_ptr, in_stride, shuffle_weight,  in cnn_convolve_no_maxpool_padding_valid_layer1_avx2()
451         const float *input_ptr = &input[k][h * in_stride];  in cnn_convolve_no_maxpool_padding_valid_layer2_avx2()  local
452         perform_convolve_for_4hx2v_2x2_blocks(input_ptr, in_stride,  in cnn_convolve_no_maxpool_padding_valid_layer2_avx2()