Classes
struct	solver_memory

Functions
__device__ void	integrate (const double t_start, const double t_end, const double pr, double __restrict__ y, const mechanism_memory __restrict__ mech, const solver_memory *__restrict__ solver)

const char *	solver_name ()
	Returns a descriptive solver name. More...

void	solver_log ()
	Executes solver specific logging tasks. More...

void	init_solver_log ()
	Initializes solver specific items for logging. More...

size_t	required_solver_size ()
	Returns the total size (in bytes) required for memory storage for a single GPU thread Used in calculation of the maximum number of possible GPU threads to launch, this method returns the size of the solver_memory structure (per-GPU thread) More...

void	createAndZero (void **ptr, size_t size)
	Convienvience method to Cuda Malloc and memset a pointer to zero. More...

void	initialize_solver (int padded, solver_memory h_mem, solver_memory d_mem)
	Initializes the GPU solver. More...

void	cleanup_solver (solver_memory h_mem, solver_memory d_mem)
	Cleans up solver memory. More...

__host__ void	check_error (int num_cond, int *codes)

Detailed Description

Include common code.

Function Documentation

◆ check_error()

__host__ void exprb43cu::check_error	(	int	num_cond,
		int *	codes
	)

Definition at line 21 of file exprb43_props.cu.

◆ cleanup_solver()

void exprb43cu::cleanup_solver	(	solver_memory **	h_mem,
		solver_memory **	d_mem
	)

Cleans up solver memory.

See also: solver_memory; solver_options.cuh

Additionally closes Krylov subspace logfiles (if LOG_OUTPUT is defined)

Definition at line 196 of file exprb43_init.cu.

◆ createAndZero()

void exprb43cu::createAndZero	(	void **	ptr,
		size_t	size
	)

Convienvience method to Cuda Malloc and memset a pointer to zero.

Parameters

ptr	The address of the pointer to malloc
size	The total size (in bytes) of the pointer to malloc

Definition at line 151 of file exprb43_init.cu.

◆ init_solver_log()

exprb43cu::init_solver_log ( )

Initializes solver specific items for logging.

Initializes the Krylov subspace logging files (if LOG_OUTPUT is defined)

See also: solver_options.cuh

Definition at line 93 of file exprb43_init.cu.

◆ initialize_solver()

void exprb43cu::initialize_solver	(	int	padded,
		solver_memory **	h_mem,
		solver_memory **	d_mem
	)

Initializes the GPU solver.

Parameters

padded	The total (padded) number of GPU threads (IVPs) to solve
h_mem	The host solver_memory structure (to be copied to the GPU)
d_mem	The device solver_memory structure (to be operated on by the GPU)

Solves for the poles and residuals used for the Rational Approximants in the Krylov subspace methods and initializes solver_memory

Definition at line 165 of file exprb43_init.cu.

◆ integrate()

__device__ void exprb43cu::integrate	(	const double	t_start,
		const double	t_end,
		const double	pr,
		double *__restrict__	y,
		const mechanism_memory *__restrict__	mech,
		const solver_memory *__restrict__	solver
	)

Definition at line 73 of file exprb43.cu.

◆ required_solver_size()

size_t exprb43cu::required_solver_size ( )

Returns the total size (in bytes) required for memory storage for a single GPU thread Used in calculation of the maximum number of possible GPU threads to launch, this method returns the size of the solver_memory structure (per-GPU thread)

See also: solver_memory

Definition at line 118 of file exprb43_init.cu.

◆ solver_log()

exprb43cu::solver_log ( )

Executes solver specific logging tasks.

Logs errors, step-sizes, and krylov subspace size (if LOG_OUTPUT is defined)

See also: solver_options.cuh

Definition at line 57 of file exprb43_init.cu.

◆ solver_name()

char * exprb43cu::solver_name ( )

Returns a descriptive solver name.

Definition at line 23 of file exprb43_init.cu.

Classes

Functions