Opm::Accelerator::openclSolverBackend< Scalar, block_size > Class Template Reference

This class implements a opencl-based ilu0-bicgstab solver on GPU. More...

#include <openclSolverBackend.hpp>

Inheritance diagram for Opm::Accelerator::openclSolverBackend< Scalar, block_size >:

Public Member Functions
	openclSolverBackend (int linear_solver_verbosity, int maxit, Scalar tolerance, unsigned int platformID, unsigned int deviceID, bool opencl_ilu_parallel, std::string linsolver)
	Construct a openclSolver. More...
	openclSolverBackend (int linear_solver_verbosity, int maxit, Scalar tolerance, bool opencl_ilu_parallel)
	For the CPR coarse solver.
SolverStatus	solve_system (std::shared_ptr< BlockedMatrix< Scalar >> matrix, Scalar *b, std::shared_ptr< BlockedMatrix< Scalar >> jacMatrix, WellContributions< Scalar > &wellContribs, GpuResult &res) override
	Solve linear system, A*x = b, matrix A must be in blocked-CSR format. More...
void	get_result (Scalar *x) override
	Solve scalar linear system, for example a coarse system of an AMG preconditioner Data is already on the GPU. More...
void	setOpencl (std::shared_ptr< cl::Context > &context, std::shared_ptr< cl::CommandQueue > &queue)
	Set OpenCL objects This class either creates them based on platformID and deviceID or receives them through this function. More...
Public Member Functions inherited from Opm::Accelerator::GpuSolver< Scalar, block_size >
	GpuSolver (int linear_solver_verbosity, int max_it, Scalar tolerance_)
	Construct a GpuSolver. More...
	GpuSolver (int linear_solver_verbosity, int max_it, Scalar tolerance_, unsigned int deviceID_)
	GpuSolver (int linear_solver_verbosity, int max_it, double tolerance_, unsigned int platformID_, unsigned int deviceID_)
virtual	~GpuSolver ()=default
	Define virtual destructor, so that the derivedclass destructor will be called.

Public Attributes
std::shared_ptr< cl::Context >	context {}
std::shared_ptr< cl::CommandQueue >	queue {}

Additional Inherited Members
Protected Attributes inherited from Opm::Accelerator::GpuSolver< Scalar, block_size >
int	verbosity = 0
int	maxit = 200
Scalar	tolerance = 1e-2
int	N
int	Nb
int	nnz
int	nnzb
unsigned int	platformID = 0
unsigned int	deviceID = 0
bool	initialized = false

Detailed Description

template<class Scalar, unsigned int block_size>
class Opm::Accelerator::openclSolverBackend< Scalar, block_size >

This class implements a opencl-based ilu0-bicgstab solver on GPU.

Constructor & Destructor Documentation

openclSolverBackend()

template<class Scalar , unsigned int block_size>

Opm::Accelerator::openclSolverBackend< Scalar, block_size >::openclSolverBackend	(	int	linear_solver_verbosity,
		int	maxit,
		Scalar	tolerance,
		unsigned int	platformID,
		unsigned int	deviceID,
		bool	opencl_ilu_parallel,
		std::string	linsolver
	)

Construct a openclSolver.

Parameters

[in]	linear_solver_verbosity	verbosity of openclSolver
[in]	maxit	maximum number of iterations for openclSolver
[in]	tolerance	required relative tolerance for openclSolver
[in]	platformID	the OpenCL platform to be used
[in]	deviceID	the device to be used
[in]	opencl_ilu_parallel	whether to parallelize the ILU decomposition and application in OpenCL with level_scheduling
[in]	linsolver	indicating the preconditioner, equal to the –linear-solver cmdline argument only ilu0, cpr_quasiimpes and isai are supported

Member Function Documentation

get_result()

template<class Scalar , unsigned int block_size>

void Opm::Accelerator::openclSolverBackend< Scalar, block_size >::get_result ( Scalar *  x )

overridevirtual

Solve scalar linear system, for example a coarse system of an AMG preconditioner Data is already on the GPU.

Get result after linear solve, and peform postprocessing if necessary

Parameters

[in,out]

x

resulting x vector, caller must guarantee that x points to a valid array

Implements Opm::Accelerator::GpuSolver< Scalar, block_size >.

setOpencl()

template<class Scalar , unsigned int block_size>

void Opm::Accelerator::openclSolverBackend< Scalar, block_size >::setOpencl	(	std::shared_ptr< cl::Context > &	context,
		std::shared_ptr< cl::CommandQueue > &	queue
	)

Set OpenCL objects This class either creates them based on platformID and deviceID or receives them through this function.

Parameters

[in]	context	the opencl context to be used
[in]	queue	the opencl queue to be used

solve_system()

template<class Scalar , unsigned int block_size>

SolverStatus Opm::Accelerator::openclSolverBackend< Scalar, block_size >::solve_system ( std::shared_ptr< BlockedMatrix< Scalar >>  matrix, Scalar *  b, std::shared_ptr< BlockedMatrix< Scalar >>  jacMatrix, WellContributions< Scalar > &  wellContribs, GpuResult &  res  )

overridevirtual

Solve linear system, A*x = b, matrix A must be in blocked-CSR format.

Parameters

[in]	matrix	matrix A
[in]	b	input vector, contains N values
[in]	jacMatrix	matrix for preconditioner
[in]	wellContribs	WellContributions, to apply them separately, instead of adding them to matrix A
[in,out]	res	summary of solver result

Returns

status code

Implements Opm::Accelerator::GpuSolver< Scalar, block_size >.

---

The documentation for this class was generated from the following files:

• opm/simulators/linalg/gpubridge/opencl/openclSolverBackend.hpp
• opm/simulators/linalg/gpubridge/opencl/openclSolverBackend.cpp