Public Types | Public Member Functions | Protected Member Functions
parallel_ode::CudaPGSSolver< CudaT, ParamsT > Class Template Reference

#include <cuda_solver.h>

Inheritance diagram for parallel_ode::CudaPGSSolver< CudaT, ParamsT >:
Inheritance graph
[legend]

List of all members.

Public Types

typedef CudaT * CudaTMutablePtr
typedef const CudaT * CudaTPtr
typedef PMemManager::mem_flags MemFlags
typedef MemManager< CudaT,
ParallelTypes::CUDA
PMemManager
typedef vec4< CudaT >::Type Vec4T

Public Member Functions

 CudaPGSSolver (int parallelFlags=DEFAULT_FLAGS, BatchType batchType=BatchTypes::DEFAULT_BATCH_TYPE, ReduceType reduceType=ReduceTypes::DEFAULT_REDUCE_TYPE, uint numBatches=ParallelOptions::MAXBATCHES)
 Constructs an instance of an CUDA-based quickstep solver.
virtual void initialize ()
 Overloaded initialization of resources specific to CUDA.
virtual ~CudaPGSSolver ()

Protected Member Functions

virtual void loadConstraints ()
virtual void preProcessDevice (const CudaT sorParam, const CudaT stepsize)
virtual void solveAndReduce (const int offset, const int batchSize)

Detailed Description

template<typename CudaT, typename ParamsT>
class parallel_ode::CudaPGSSolver< CudaT, ParamsT >

Definition at line 10 of file cuda_solver.h.


Member Typedef Documentation

template<typename CudaT , typename ParamsT >
typedef CudaT* parallel_ode::CudaPGSSolver< CudaT, ParamsT >::CudaTMutablePtr
template<typename CudaT , typename ParamsT >
typedef const CudaT* parallel_ode::CudaPGSSolver< CudaT, ParamsT >::CudaTPtr
template<typename CudaT , typename ParamsT >
typedef PMemManager::mem_flags parallel_ode::CudaPGSSolver< CudaT, ParamsT >::MemFlags
template<typename CudaT , typename ParamsT >
typedef MemManager<CudaT,ParallelTypes::CUDA> parallel_ode::CudaPGSSolver< CudaT, ParamsT >::PMemManager
template<typename CudaT , typename ParamsT >
typedef vec4<CudaT>::Type parallel_ode::CudaPGSSolver< CudaT, ParamsT >::Vec4T

Constructor & Destructor Documentation

template<typename CudaT , typename ParamsT >
parallel_ode::CudaPGSSolver< CudaT, ParamsT >::CudaPGSSolver ( int  parallelFlags = DEFAULT_FLAGS,
BatchType  batchType = BatchTypes::DEFAULT_BATCH_TYPE,
ReduceType  reduceType = ReduceTypes::DEFAULT_REDUCE_TYPE,
uint  numBatches = ParallelOptions::MAXBATCHES 
) [inline]

Constructs an instance of an CUDA-based quickstep solver.

Parameters:
parallelFlagsFlags controller various parallel functionality
batchTypeThe type of batch strategy to be used
reduceTypeThe Type of reduction to be used
numBatchesThe maximum number of batches to be used, if any

Definition at line 27 of file cuda_solver.h.

template<typename CudaT , typename ParamsT >
virtual parallel_ode::CudaPGSSolver< CudaT, ParamsT >::~CudaPGSSolver ( ) [inline, virtual]

Definition at line 36 of file cuda_solver.h.


Member Function Documentation

template<typename CudaT , typename ParamsT >
void parallel_ode::CudaPGSSolver< CudaT, ParamsT >::initialize ( ) [virtual]

Overloaded initialization of resources specific to CUDA.

Reimplemented from parallel_ode::ParallelPGSSolver< CudaT, ParamsT, ParallelTypes::CUDA >.

Definition at line 17 of file cuda_solver.cpp.

template<typename CudaT , typename ParamsT >
void parallel_ode::CudaPGSSolver< CudaT, ParamsT >::loadConstraints ( ) [protected, virtual]
template<typename CudaT , typename ParamsT >
void parallel_ode::CudaPGSSolver< CudaT, ParamsT >::preProcessDevice ( const CudaT  sorParam,
const CudaT  stepsize 
) [protected, virtual]
Todo:
Finish preprocessing implementation

Reimplemented from parallel_ode::ParallelPGSSolver< CudaT, ParamsT, ParallelTypes::CUDA >.

Definition at line 28 of file cuda_solver.cpp.

template<typename CudaT , typename ParamsT >
void parallel_ode::CudaPGSSolver< CudaT, ParamsT >::solveAndReduce ( const int  offset,
const int  batchSize 
) [protected, virtual]

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


parallel_quickstep
Author(s): Jared Duke
autogenerated on Fri Jan 3 2014 11:36:56