Halide 19.0.0
Halide compiler and libraries
|
#include <cstdint>
#include <vector>
Go to the source code of this file.
Namespaces | |
namespace | Halide |
This file defines the class FunctionDAG, which is our representation of a Halide pipeline, and contains methods to using Halide's bounds tools to query properties of it. | |
namespace | Halide::Internal |
namespace | Halide::Internal::Autoscheduler |
Functions | |
bool | Halide::Internal::Autoscheduler::all_ones (const std::vector< int64_t > &nums) |
bool | Halide::Internal::Autoscheduler::equal_to_existing_size (const std::vector< int64_t > &s, const std::vector< int64_t > &nums) |
std::vector< std::vector< int64_t > > | Halide::Internal::Autoscheduler::generate_serial_tilings (const std::vector< int64_t > &s, int d, int last_d, int vectorized_index, const std::vector< int > &vec_dim_serial_sizes, bool filter_small_outer_extents=false, bool allow_inner_ones=false) |
std::vector< std::vector< int64_t > > | Halide::Internal::Autoscheduler::generate_tilings (const std::vector< int64_t > &s, int d, int factor, bool allow_splits, const std::vector< int > &inner_sizes=std::vector< int >()) |
void | Halide::Internal::Autoscheduler::lowered_dims (const std::vector< int64_t > &size, int vector_loop_i, std::vector< int64_t > &lowered_size) |
moves vectorized dimension first and also removes dimensions with size 1 to reflect actual thread dimensions when loop nests are lowered | |
std::vector< std::vector< int64_t > > | Halide::Internal::Autoscheduler::generate_gpu_tilings (const std::vector< std::vector< int64_t > > &stage_sizes, const std::vector< std::vector< int > > &pure_dims, const std::vector< int64_t > &max_s, int d, const std::vector< int > &vectorized_indices, bool serial_inner, bool is_compute_root_stage) |