Public Attributes
int	blocks
	CUDA blocks.

int	nbodies
	each node corresponds to a split of the cubic box in 3D space to equal cubic boxes hence, 8 octant nodes per particle is a theoretical octree limit: a maximal number of octree nodes is "nnodesd" and a number of particles "nbodiesd" respectively.

int	nnodes

float *	r
	particle positions on the device:

float *	u
	particle dipole moments on the device:

float *	mass
	Not a real mass. Just a node weight coefficient.

float *	minp
	min positions' coordinates of the Barnes-Hut box.

float *	maxp
	max positions' coordinates of the Barnes-Hut box.

int *	err
	Error report.

int *	sort
	Indices of particles sorted according to the tree linear representation.

int *	child
	The tree linear representation.

int *	count
	Supplementary array: a tree nodes (division octant cells/particles inside) counting.

int *	start
	Start indices for the per-cell sorting.

int *	max_lps
	trace the max loops for a threads' sync

Detailed Description

Definition at line 27 of file barnes_hut_gpu_cuda.cuh.

Member Data Documentation

◆ blocks

int BHData::blocks

CUDA blocks.

Definition at line 29 of file barnes_hut_gpu_cuda.cuh.

Referenced by allocBHmemCopy(), energyBH(), and forceBH().

◆ child

int* BHData::child

The tree linear representation.

Definition at line 51 of file barnes_hut_gpu_cuda.cuh.

Referenced by __launch_bounds__(), __launch_bounds__(), __launch_bounds__(), __launch_bounds__(), allocBHmemCopy(), deallocBH(), and if().

◆ count

int* BHData::count

Supplementary array: a tree nodes (division octant cells/particles inside) counting.

Definition at line 54 of file barnes_hut_gpu_cuda.cuh.

Referenced by __launch_bounds__(), __launch_bounds__(), allocBHmemCopy(), and deallocBH().

◆ err

int* BHData::err

Error report.

Definition at line 47 of file barnes_hut_gpu_cuda.cuh.

Referenced by __launch_bounds__(), allocBHmemCopy(), deallocBH(), energyBH(), forceBH(), if(), and initializationKernel().

◆ mass

float* BHData::mass

Not a real mass. Just a node weight coefficient.

Definition at line 41 of file barnes_hut_gpu_cuda.cuh.

Referenced by __launch_bounds__(), __launch_bounds__(), __launch_bounds__(), allocBHmemCopy(), and deallocBH().

◆ max_lps

int* BHData::max_lps

trace the max loops for a threads' sync

Definition at line 58 of file barnes_hut_gpu_cuda.cuh.

Referenced by __launch_bounds__(), __launch_bounds__(), __launch_bounds__(), allocBHmemCopy(), deallocBH(), and initializationKernel().

◆ maxp

float* BHData::maxp

max positions' coordinates of the Barnes-Hut box.

Definition at line 45 of file barnes_hut_gpu_cuda.cuh.

Referenced by __launch_bounds__(), allocBHmemCopy(), and deallocBH().

◆ minp

float* BHData::minp

min positions' coordinates of the Barnes-Hut box.

Definition at line 43 of file barnes_hut_gpu_cuda.cuh.

Referenced by __launch_bounds__(), allocBHmemCopy(), and deallocBH().

◆ nbodies

int BHData::nbodies

each node corresponds to a split of the cubic box in 3D space to equal cubic boxes hence, 8 octant nodes per particle is a theoretical octree limit: a maximal number of octree nodes is "nnodesd" and a number of particles "nbodiesd" respectively.

Definition at line 34 of file barnes_hut_gpu_cuda.cuh.

Referenced by __launch_bounds__(), __launch_bounds__(), __launch_bounds__(), __launch_bounds__(), allocBHmemCopy(), fill_bh_data(), and if().