Searched defs:num_tiles (Results 1 - 6 of 6) sorted by relevance
/external/tensorflow/tensorflow/compiler/xla/client/ |
H A D | sharding_builder.cc | 49 OpSharding Tile1D(const Shape& tile_shape, int64 num_tiles) { argument 54 std::vector<int64> dimensions(1, num_tiles); 58 tile_dimension = CeilOfRatio(static_cast<int64>(tile_dimension), num_tiles); 59 result.add_tile_assignment_dimensions(num_tiles); 60 for (int64 i = 0; i < num_tiles; ++i) {
|
/external/mesa3d/src/gallium/drivers/freedreno/ |
H A D | freedreno_query_hw.h | 111 uint32_t num_tiles; member in struct:fd_hw_sample 146 void fd_hw_query_prepare(struct fd_batch *batch, uint32_t num_tiles);
|
H A D | freedreno_query_hw.c | 278 assert(start->num_tiles == end->num_tiles); 293 for (i = 0; i < start->num_tiles; i++) { 350 samp->num_tiles = 0; 386 fd_hw_query_prepare(struct fd_batch *batch, uint32_t num_tiles) argument 391 fd_resource_resize(batch->query_buf, tile_stride * num_tiles); 398 samp->num_tiles = num_tiles;
|
/external/tensorflow/tensorflow/compiler/xla/service/ |
H A D | hlo_sharding.cc | 29 HloSharding HloSharding::Tile1D(const Shape& input_shape, int64 num_tiles) { argument 31 CHECK_GT(num_tiles, 1); 32 std::vector<int64> dimensions(1, num_tiles); 35 tile_dimension = CeilOfRatio(static_cast<int64>(tile_dimension), num_tiles);
|
/external/tensorflow/tensorflow/compiler/xla/service/gpu/ |
H A D | ir_emitter_unnested.cc | 822 int64 num_tiles = local 828 0, num_tiles, 896 return num_tiles; 923 int64 num_tiles = EmitTranspose021Tiled( local 929 UpdateLaunchDimensions(LaunchDimensions(num_tiles, num_rows * tile_size), 949 int64 num_tiles = local 955 bool all_threads_in_bounds = num_tiles * kTileSize == num_elems; 1112 reduce->shape().element_type(), {num_tiles}, {0});
|
/external/tensorflow/tensorflow/core/kernels/ |
H A D | deep_conv2d.cc | 67 const int64 num_tiles = row_tiles * col_tiles; local 70 return num_tiles * 566 // [num_tiles, in_depth] 569 // [num_tiles, out_depth, shard_rows, shard_cols] 625 // [tile_rows, tile_cols, num_tiles, in_depth] 631 const int64 num_tiles, const int64 in_r_start, 638 const int64 coord_stride = num_tiles * args.in_depth; 670 // Transforms 'num_tiles' tiles from 'input' by 'transform_matrix', storing the 677 // [tile_rows, tile_cols, num_tiles, in_depth] 681 // [tile_rows, tile_cols, num_tiles, in_dept 629 operator ()(const Conv2DArgs& args, const DeepConv2DTransform<T>* transform, const int64 num_tiles, const int64 in_r_start, const int64 in_c_start, const T* input, T* tile_buffer) argument 692 operator ()(const Conv2DArgs& args, const DeepConv2DTransform<T>* transform, const int64 num_tiles, const int64 in_r_start, const int64 in_c_start, const T* input, const T* transform_matrix, T* tile_buffer, T* tile_transform) argument 744 operator ()(const Conv2DArgs& args, const DeepConv2DTransform<T>* transform, const int64 num_tiles, const int64 in_r, const int64 in_c, const int64 filter_shards_row, const int64 filter_shards_col, const T* out_transform_matrix, const T* out_buffer, T* out_transform_buffer, T* output) argument 878 operator ()(const Conv2DArgs& args, const DeepConv2DTransform<T>* transform, const Conv2DState<T>& cs, const int64 in_r, const int64 in_c, const int64 num_tiles, const std::vector<Tensor>& packed_filters, const T* input, T* output) argument [all...] |
Completed in 274 milliseconds