Opm::cuistl::CuOwnerOverlapCopy< field_type, block_size, OwnerOverlapCopyCommunicationType > Class Template Reference

CUDA compatiable variant of Dune::OwnerOverlapCopyCommunication. More...

#include <CuOwnerOverlapCopy.hpp>

Public Types
using	X = CuVector< field_type >

Public Member Functions
	CuOwnerOverlapCopy (const OwnerOverlapCopyCommunicationType &cpuOwnerOverlapCopy)
void	dot (const X &x, const X &y, field_type &output) const
	dot will carry out the dot product between x and y on the owned indices, then sum up the result across MPI processes.
field_type	norm (const X &x) const
	norm computes the l^2-norm of x across processes.
void	project (X &x) const
	project will project x to the owned subspace
void	copyOwnerToAll (const X &source, X &dest) const
	copyOwnerToAll will copy source to the CPU, then call OwnerOverlapCopyCommunicationType::copyOwnerToAll on the copied data, and copy the result back to the GPU

Detailed Description

template<class field_type, int block_size, class OwnerOverlapCopyCommunicationType>
class Opm::cuistl::CuOwnerOverlapCopy< field_type, block_size, OwnerOverlapCopyCommunicationType >

CUDA compatiable variant of Dune::OwnerOverlapCopyCommunication.

This class can essentially be seen as an adapter around Dune::OwnerOverlapCopyCommunication, and should work as a Dune::OwnerOverlapCopyCommunication on CuVectors

Note

This currently only has the functionality to parallelize the linear solve.

Template Parameters

field_type	should be a field_type supported by CuVector (double, float)
block_size	the block size used (this is relevant for say figuring out the correct indices)
OwnerOverlapCopyCommunicationType	should mimic Dune::OwnerOverlapCopyCommunication.

Member Function Documentation

copyOwnerToAll()

template<class field_type , int block_size, class OwnerOverlapCopyCommunicationType >

void Opm::cuistl::CuOwnerOverlapCopy< field_type, block_size, OwnerOverlapCopyCommunicationType >::copyOwnerToAll ( const X &  source, X &  dest  ) const

inline

copyOwnerToAll will copy source to the CPU, then call OwnerOverlapCopyCommunicationType::copyOwnerToAll on the copied data, and copy the result back to the GPU

Parameters

[in]	source
[out]	dest

dot()

template<class field_type , int block_size, class OwnerOverlapCopyCommunicationType >

void Opm::cuistl::CuOwnerOverlapCopy< field_type, block_size, OwnerOverlapCopyCommunicationType >::dot ( const X &  x, const X &  y, field_type &  output  ) const

inline

dot will carry out the dot product between x and y on the owned indices, then sum up the result across MPI processes.

Parameters

[out]

output

result will be stored here

Note

This uses the same interface as its DUNE equivalent.

norm()

template<class field_type , int block_size, class OwnerOverlapCopyCommunicationType >

field_type Opm::cuistl::CuOwnerOverlapCopy< field_type, block_size, OwnerOverlapCopyCommunicationType >::norm ( const X &  x ) const

inline

norm computes the l^2-norm of x across processes.

This will compute the dot product of x with itself on owned indices, then sum the result across process and return the square root of the sum.

project()

template<class field_type , int block_size, class OwnerOverlapCopyCommunicationType >

void Opm::cuistl::CuOwnerOverlapCopy< field_type, block_size, OwnerOverlapCopyCommunicationType >::project ( X &  x ) const

inline

project will project x to the owned subspace

For each component i which is not owned, x_i will be set to 0

Parameters

[in,out]

x

the vector to project

---

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

• opm/simulators/linalg/cuistl/CuOwnerOverlapCopy.hpp