cpu_trans_map_amr.hpp File Reference

#include <vector>
#include "vec.h"
#include "../common.h"
#include "../spatial_cells/spatial_cell_wrapper.hpp"

Include dependency graph for cpu_trans_map_amr.hpp:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Functions
bool	trans_map_1d_amr (const dccrg::Dccrg< spatial_cell::SpatialCell, dccrg::Cartesian_Geometry > &mpiGrid, const std::vector< CellID > &localPropagatedCells, const std::vector< CellID > &remoteTargetCells, std::vector< uint > &nPencils, const uint dimension, const Realf dt, const uint popID)
void	update_remote_mapping_contribution_amr (dccrg::Dccrg< spatial_cell::SpatialCell, dccrg::Cartesian_Geometry > &mpiGrid, const uint dimension, int direction, const uint popID)

Function Documentation

trans_map_1d_amr()

bool trans_map_1d_amr	(	const dccrg::Dccrg< spatial_cell::SpatialCell, dccrg::Cartesian_Geometry > &	mpiGrid,
		const std::vector< CellID > &	localPropagatedCells,
		const std::vector< CellID > &	remoteTargetCells,
		std::vector< uint > &	nPencils,
		const uint	dimension,
		const Realf	dt,
		const uint	popID )

update_remote_mapping_contribution_amr()

void update_remote_mapping_contribution_amr	(	dccrg::Dccrg< spatial_cell::SpatialCell, dccrg::Cartesian_Geometry > &	mpiGrid,
		const uint	dimension,
		int	direction,
		const uint	popID )

GPUTODO: First attempts at using unified memory for remote neighbours Should probably: a) Switch to using pure device buffers. Some MPI-CUDA implementations do not work well with UB buffers. b) transition to re-using buffers and ensuring size is suitable? If that path is taken, it should also check for any local cells not on process boundary (anymore, due to LB) and free the buffers from those cells... so gets tricky.

Definition at line 493 of file cpu_trans_map_amr.cpp.