/external/pthreadpool/src/ |
D | threadpool-legacy.c | 11 static inline size_t divide_round_up(size_t dividend, size_t divisor) { in divide_round_up() function 124 const size_t tile_range_i = divide_round_up(range_i, tile_i); in pthreadpool_compute_3d_tiled() 125 const size_t tile_range_j = divide_round_up(range_j, tile_j); in pthreadpool_compute_3d_tiled() 126 const size_t tile_range_k = divide_round_up(range_k, tile_k); in pthreadpool_compute_3d_tiled() 211 const size_t tile_range_i = divide_round_up(range_i, tile_i); in pthreadpool_compute_4d_tiled() 212 const size_t tile_range_j = divide_round_up(range_j, tile_j); in pthreadpool_compute_4d_tiled() 213 const size_t tile_range_k = divide_round_up(range_k, tile_k); in pthreadpool_compute_4d_tiled() 214 const size_t tile_range_l = divide_round_up(range_l, tile_l); in pthreadpool_compute_4d_tiled()
|
D | threadpool-pthreads.c | 72 static inline size_t divide_round_up(size_t dividend, size_t divisor) { in divide_round_up() function 635 const size_t tile_range = divide_round_up(range, tile); in pthreadpool_parallelize_1d_tile_1d() 737 const size_t tile_range_j = divide_round_up(range_j, tile_j); in pthreadpool_parallelize_2d_tile_1d() 799 const size_t tile_range_i = divide_round_up(range_i, tile_i); in pthreadpool_parallelize_2d_tile_2d() 800 const size_t tile_range_j = divide_round_up(range_j, tile_j); in pthreadpool_parallelize_2d_tile_2d() 870 const size_t tile_range_j = divide_round_up(range_j, tile_j); in pthreadpool_parallelize_3d_tile_2d() 871 const size_t tile_range_k = divide_round_up(range_k, tile_k); in pthreadpool_parallelize_3d_tile_2d() 952 const size_t tile_range_k = divide_round_up(range_k, tile_k); in pthreadpool_parallelize_4d_tile_2d() 953 const size_t tile_range_l = divide_round_up(range_l, tile_l); in pthreadpool_parallelize_4d_tile_2d() 1043 const size_t tile_range_l = divide_round_up(range_l, tile_l); in pthreadpool_parallelize_5d_tile_2d() [all …]
|
/external/XNNPACK/src/ |
D | deconvolution-nhwc.c | 216 const size_t subkernel_height = divide_round_up(kernel_height - offset_y, stride_height); in xnn_create_deconvolution2d_nhwc_q8() 217 const size_t subkernel_width = divide_round_up(kernel_width - offset_x, stride_width); in xnn_create_deconvolution2d_nhwc_q8() 465 const size_t subkernel_height = divide_round_up(kernel_height - offset_y, stride_height); in xnn_create_deconvolution2d_nhwc_f32() 466 const size_t subkernel_width = divide_round_up(kernel_width - offset_x, stride_width); in xnn_create_deconvolution2d_nhwc_f32() 622 const size_t num_other_tiles = groups * batch_size * divide_round_up(output_size, mr); in setup_conv_path() 624 …const size_t max_nc = divide_round_up(group_output_channels * num_other_tiles, num_threads * targe… in setup_conv_path() 626 nc = min(nc, divide_round_up(nc, max_nc * nr) * nr); in setup_conv_path() 679 …sizeof(void*) * kernel_size * output_height * stride_width * round_up(divide_round_up(output_width… in setup_subconv2d_path() 706 … subconvolution_params->slice_width = divide_round_up(output_width - output_x_start, stride_width); in setup_subconv2d_path() 707 …subconvolution_params->slice_height = divide_round_up(output_height - output_y_start, stride_heigh… in setup_subconv2d_path() [all …]
|
D | prelu-nc.c | 143 … const size_t max_batch_tile = divide_round_up(batch_size, num_threads * target_tiles_per_thread); in xnn_setup_prelu_nc_f32() 146 … batch_tile = min(batch_tile, divide_round_up(batch_tile, max_batch_tile * row_tile) * row_tile); in xnn_setup_prelu_nc_f32()
|
D | convolution-nhwc.c | 44 return divide_round_up(input_dimension, subsampling_dimension); in compute_output_dimension_with_tf_same_padding() 828 const size_t num_other_tiles = groups * divide_round_up(batch_output_size, mr); in setup_convolution2d_nhwc() 830 …const size_t max_nc = divide_round_up(group_output_channels * num_other_tiles, num_threads * targe… in setup_convolution2d_nhwc() 832 nc = min(nc, divide_round_up(nc, max_nc * nr) * nr); in setup_convolution2d_nhwc() 919 const size_t num_other_tiles = groups * batch_size * divide_round_up(output_size, mr); in setup_convolution2d_nhwc() 921 …const size_t max_nc = divide_round_up(group_output_channels * num_other_tiles, num_threads * targe… in setup_convolution2d_nhwc() 923 nc = min(nc, divide_round_up(nc, max_nc * nr) * nr); in setup_convolution2d_nhwc() 1028 … const size_t max_mc = divide_round_up(batch_output_size, num_threads * target_tiles_per_thread); in setup_convolution2d_nhwc() 1031 mc = min(mc, divide_round_up(mc, max_mc * mr) * mr); in setup_convolution2d_nhwc()
|
D | resize-bilinear-nhwc.c | 206 …const size_t max_output_size_tile = divide_round_up(output_size, num_threads * target_tiles_per_th… in xnn_setup_resize_bilinear2d_nhwc_f32() 211 …divide_round_up(output_size_tile, max_output_size_tile * output_size_subtile) * output_size_subtil… in xnn_setup_resize_bilinear2d_nhwc_f32()
|
D | fully-connected-nc.c | 391 const size_t num_other_tiles = divide_round_up(batch_size, mr); in setup_fully_connected_nc() 393 …const size_t max_nc = divide_round_up(output_channels * num_other_tiles, num_threads * target_tile… in setup_fully_connected_nc() 395 nc = min(nc, divide_round_up(nc, max_nc * nr) * nr); in setup_fully_connected_nc()
|
D | convolution-nchw.c | 594 const size_t max_mc = divide_round_up(input_size, num_threads * target_tiles_per_thread); in setup_convolution2d_nchw() 596 mc = min(mc, divide_round_up(mc, max_mc * mr) * mr); in setup_convolution2d_nchw() 640 …const size_t max_output_height_slice = divide_round_up(output_height, num_threads * target_tiles_p… in setup_convolution2d_nchw() 643 …divide_round_up(output_height_slice, max_output_height_slice * output_height_tile) * output_height… in setup_convolution2d_nchw()
|
D | indirection.c | 215 … const size_t sliced_output_width = divide_round_up(output_width - output_x_start, stride_width); in xnn_indirection_init_subconv2d()
|
/external/XNNPACK/src/xnnpack/ |
D | math.h | 30 inline static size_t divide_round_up(size_t n, size_t q) { in divide_round_up() function 35 return divide_round_up(n, q) * q; in round_up()
|
D | pack.h | 243 …const int32_t boff = (int32_t) divide_round_up(kh - oy, sh) * (int32_t) divide_round_up(kw - ox, s… in xnn_pack_q8_deconv_goki_w()
|