Functions
__global__ void	intDriver (const int NUM, const double t, const double t_end, const double __restrict__ pr_global, double __restrict__ y_global, const mechanism_memory __restrict__ d_mem, const solver_memory __restrict__ s_mem)
	Generic driver for the GPU integrators. More...

__device__ void	integrate (const double, const double, const double, double const __restrict__, mechanism_memory const const __restrict__, solver_memory const *const __restrict__)

__host__ void	check_error (int num_conditions, int *code_arr)

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

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

void	initialize_solver (const int, solver_memory , solver_memory )

void	cleanup_solver (solver_memory , solver_memory )

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

void	memcpy2D_in (double dst, const int pitch_dst, double const src, const int pitch_src, const int offset, const size_t width, const int height)
	A convienience method to copy memory between host pointers of different pitches, widths and heights. Enables easier use of CUDA's cudaMemcpy2D functions. More...

void	memcpy2D_out (double dst, const int pitch_dst, double const src, const int pitch_src, const int offset, const size_t width, const int height)
	A convienience method to copy memory between host pointers of different pitches, widths and heights. Enables easier use of CUDA's cudaMemcpy2D functions. More...

void	accelerInt_initialize (int NUM, int device)
	Initializes the solver. More...

void	accelerInt_integrate (const int NUM, const double t_start, const double t_end, const double stepsize, double __restrict__ y_host, const double __restrict__ var_host)
	integrate NUM odes from time `t_start` to time `t_end`, using stepsizes of `stepsize` More...

void	accelerInt_cleanup ()
	Cleans up the solver. More...

Variables
int	padded
	Padded # of ODEs to solve. More...

solver_memory *	host_solver
	The solver memory structs. More...

solver_memory *	device_solver

mechanism_memory *	host_mech
	The mechanism memory structs. More...

mechanism_memory *	device_mech

dim3	dimBlock
	block and grid sizes More...

dim3	dimGrid

int *	result_flag
	result flag More...

double *	y_temp
	temorary storage More...

Function Documentation

◆ accelerInt_cleanup()

void genericcu::accelerInt_cleanup ( )

Cleans up the solver.

Definition at line 217 of file solver_interface.cu.

◆ accelerInt_initialize()

void genericcu::accelerInt_initialize	(	int	NUM,
		int	device
	)

Initializes the solver.

Parameters

[in]	NUM	The number of ODEs to integrate
[in]	device	The CUDA device number, if < 0 set to the first available GPU

Definition at line 88 of file solver_interface.cu.

◆ accelerInt_integrate()

void genericcu::accelerInt_integrate	(	const int	NUM,
		const double	t_start,
		const double	t_end,
		const double	stepsize,
		double *__restrict__	y_host,
		const double *__restrict__	var_host
	)

integrate NUM odes from time t_start to time t_end, using stepsizes of stepsize

integrate NUM odes from time t_start to time t_end, using stepsizes of t_step

Parameters

[in] NUM The number of ODEs to integrate. This should be the size of the leading dimension of y_host and var_host.

See also: accelerint_indx

Parameters

[in]	t_start	The starting time
[in]	t_end	The end time
[in]	stepsize	The integration step size. If `stepsize` < 0, the step size will be set to `t_end - t`
[in,out]	y_host	The state vectors to integrate.
[in]	var_host	The parameters to use in dydt() and eval_jacob()
[in]	NUM	The number of ODEs to integrate. This should be the size of the leading dimension of `y_host` and `var_host`.

See also: accelerint_indx

Parameters

[in]	t_start	The system time
[in]	t_end	The end time
[in]	stepsize	The integration step size. If `stepsize` < 0, the step size will be set to `t_end - t`
[in,out]	y_host	The state vectors to integrate.
[in]	var_host	The parameters to use in dydt() and eval_jacob()

Definition at line 161 of file solver_interface.cu.

◆ check_error()

__host__ void genericcu::check_error	(	int	num_conditions,
		int *	code_arr
	)

Definition at line 15 of file radau2a_props.cu.

◆ cleanup_solver()

void genericcu::cleanup_solver	(	solver_memory **	,
		solver_memory **
	)

◆ init_solver_log()

void genericcu::init_solver_log ( )

Initializes solver specific items for logging.

Initializes stepsize logging for stiffness measurement

Initializes solver specific items for logging.

See also: solver_options.cuh

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

See also: solver_options.cuh

Definition at line 190 of file cvodes_init.c.

◆ initialize_solver()

void genericcu::initialize_solver	(	const int	,
		solver_memory **	,
		solver_memory **
	)

◆ intDriver()

__global__ void genericcu::intDriver	(	const int	NUM,
		const double	t,
		const double	t_end,
		const double *__restrict__	pr_global,
		double *__restrict__	y_global,
		const mechanism_memory *__restrict__	d_mem,
		const solver_memory *__restrict__	s_mem
	)

Generic driver for the GPU integrators.

Parameters

[in]	NUM	The (non-padded) number of IVPs to integrate
[in]	t	The current system time
[in]	t_end	The IVP integration end time
[in]	pr_global	The system constant variable (pressures / densities)
[in,out]	y_global	The system state vectors at time t. Returns system state vectors at time t_end
[in]	d_mem	The mechanism_memory struct that contains the pre-allocated memory for the RHS \ Jacobian evaluation
[in]	s_mem	The solver_memory struct that contains the pre-allocated memory for the solver

Definition at line 29 of file solver_generic.cu.

◆ integrate()

__device__ void genericcu::integrate	(	const double	,
		const double	,
		const double	,
		double *const	__restrict__,
		mechanism_memory const *const	__restrict__,
		solver_memory const *const	__restrict__
	)

◆ memcpy2D_in()

void genericcu::memcpy2D_in	(	double *	dst,
		const int	pitch_dst,
		double const *	src,
		const int	pitch_src,
		const int	offset,
		const size_t	width,
		const int	height
	)

inline

A convienience method to copy memory between host pointers of different pitches, widths and heights. Enables easier use of CUDA's cudaMemcpy2D functions.

Parameters

[out]	dst	The destination array
[in]	pitch_dst	The width (in bytes) of the destination array. This corresponds to the padded number of IVPs to be solved.
[in]	src	The source pointer
[in]	pitch_src	The width (in bytes) of the source array. This corresponds to the (non-padded) number of IVPs read by read_initial_conditions
[in]	offset	The offset within the source array (IVP index) to copy from. This is useful in the case (for large models) where the solver and state vector memory will not fit in device memory and the integration must be split into multiple kernel calls.
[in]	width	The size (in bytes) of memory to copy for each entry in the state vector
[in]	height	The number of entries in the state vector

Definition at line 46 of file solver_interface.cu.

◆ memcpy2D_out()

void genericcu::memcpy2D_out	(	double *	dst,
		const int	pitch_dst,
		double const *	src,
		const int	pitch_src,
		const int	offset,
		const size_t	width,
		const int	height
	)

inline

A convienience method to copy memory between host pointers of different pitches, widths and heights. Enables easier use of CUDA's cudaMemcpy2D functions.

Parameters

[out]	dst	The destination array
[in]	pitch_dst	The width (in bytes) of the source array. This corresponds to the (non-padded) number of IVPs read by read_initial_conditions
[in]	src	The source pointer
[in]	pitch_src	The width (in bytes) of the destination array. This corresponds to the padded number of IVPs to be solved.
[in]	offset	The offset within the destination array (IVP index) to copy to. This is useful in the case (for large models) where the solver and state vector memory will not fit in device memory and the integration must be split into multiple kernel calls.
[in]	width	The size (in bytes) of memory to copy for each entry in the state vector
[in]	height	The number of entries in the state vector

Definition at line 72 of file solver_interface.cu.

◆ required_solver_size()

size_t genericcu::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 167 of file exp4_init.cu.

◆ solver_log()

void genericcu::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 192 of file cvodes_init.c.

◆ solver_name()

const char* genericcu::solver_name ( )

Returns a descriptive solver name.

Definition at line 181 of file cvodes_init.c.

Variable Documentation

◆ device_mech

mechanism_memory * genericcu::device_mech

Definition at line 22 of file solver_interface.cu.

◆ device_solver

solver_memory * genericcu::device_solver

Definition at line 20 of file solver_interface.cu.

◆ dimBlock

dim3 genericcu::dimBlock

block and grid sizes

Definition at line 24 of file solver_interface.cu.

◆ dimGrid

dim3 genericcu::dimGrid

Definition at line 24 of file solver_interface.cu.

◆ host_mech

mechanism_memory* genericcu::host_mech

The mechanism memory structs.

Definition at line 22 of file solver_interface.cu.

◆ host_solver

solver_memory* genericcu::host_solver

The solver memory structs.

Definition at line 20 of file solver_interface.cu.

◆ padded

int genericcu::padded

Padded # of ODEs to solve.

Definition at line 18 of file solver_interface.cu.

◆ result_flag

int* genericcu::result_flag

result flag

Definition at line 26 of file solver_interface.cu.

◆ y_temp

double* genericcu::y_temp

temorary storage

Definition at line 28 of file solver_interface.cu.

Functions

Variables

Function Documentation

◆ accelerInt_cleanup()

◆ accelerInt_initialize()

◆ accelerInt_integrate()

◆ check_error()

◆ cleanup_solver()

◆ init_solver_log()

◆ initialize_solver()

◆ intDriver()

◆ integrate()

◆ memcpy2D_in()

◆ memcpy2D_out()

◆ required_solver_size()

◆ solver_log()

◆ solver_name()

Variable Documentation

◆ device_mech

◆ device_solver

◆ dimBlock

◆ dimGrid

◆ host_mech

◆ host_solver

◆ padded

◆ result_flag

◆ y_temp