pzgssvx3d_csc_batch.c File Reference

#include "superlu_ddefs.h"
#include "superlu_defs.h"
#include "superlu_upacked.h"
#include <stdbool.h>

Include dependency graph for pzgssvx3d_csc_batch.c:

Functions
int	file_zPrint_CompRowLoc_to_Triples (SuperMatrix *A)
int	pzgssvx3d_csc_batch (superlu_dist_options_t *options, int batchCount, int m, int n, int nnz, int nrhs, handle_t *SparseMatrix_handles, doublecomplex **RHSptr, int *ldRHS, double **ReqPtr, double **CeqPtr, int **RpivPtr, int **CpivPtr, DiagScale_t *DiagScale, handle_t *F, doublecomplex **Xptr, int *ldX, double **Berrs, gridinfo3d_t *grid3d, SuperLUStat_t *stat, int *info)
	Solve a batch of linear systems Ai * Xi = Bi with direct method, computing the LU factorization of each matrix Ai; This is the fixed-size interface: all the input matrices have the same sparsity structure. More...

Detailed Description

Copyright (c) 2003, The Regents of the University of California, through Lawrence Berkeley National Laboratory (subject to receipt of any required approvals from U.S. Dept. of Energy)

All rights reserved.

The source code is distributed under BSD license, see the file License.txt at the top-level directory.

Function Documentation

file_zPrint_CompRowLoc_to_Triples()

int file_zPrint_CompRowLoc_to_Triples

(

SuperMatrix *

A

)

pzgssvx3d_csc_batch()

int pzgssvx3d_csc_batch	(	superlu_dist_options_t *	options,
		int	batchCount,
		int	m,
		int	n,
		int	nnz,
		int	nrhs,
		handle_t *	SparseMatrix_handles,
		doublecomplex **	RHSptr,
		int *	ldRHS,
		double **	ReqPtr,
		double **	CeqPtr,
		int **	RpivPtr,
		int **	CpivPtr,
		DiagScale_t *	DiagScale,
		handle_t *	F,
		doublecomplex **	Xptr,
		int *	ldX,
		double **	Berrs,
		gridinfo3d_t *	grid3d,
		SuperLUStat_t *	stat,
		int *	info
	)

Solve a batch of linear systems Ai * Xi = Bi with direct method, computing the LU factorization of each matrix Ai;
This is the fixed-size interface: all the input matrices have the same sparsity structure.

Parameters

  

    
[in]
options
solver options

    
[in]
batchCount
number of matrices in the batch

    
[in]
m
row dimension of the matrices

    
[in]
n
column dimension of the matrices

    
[in]
nnz
number of non-zero entries in each matrix

    
[in]
nrhs
number of right-hand-sides

    
[in,out]
SparseMatrix_handles
 array of sparse matrix handles, of size 'batchCount', each pointing to the actual storage in CSC format, see 'NCformat' in SuperMatix structure
     Each A is overwritten by row/col scaling R*A*C

    
[in,out]
RHSptr
 array of pointers to dense storage of right-hand sides B
     Each B is overwritten by row/col scaling R*B*C

    
[in]
ldRHS
array of leading dimensions of RHS

    
[in,out]
ReqPtr
array of pointers to diagonal row scaling vectors R, each of size m
   ReqPtr[] are allocated internally if equilibration is asked for

    
[in,out]
CeqPtr
array of pointers to diagonal colum scaling vectors C, each of size n
   CeqPtr[] are allocated internally if equilibration is asked for

    
[in,out]
RpivPtr
array of pointers to row permutation vectors, each of size m

    
[in,out]
CpivPtr
array of pointers to column permutation vectors, each of size n

    
[in,out]
DiagScale
array of indicators how equilibration is done for each matrix

    
[out]
F
array of handles pointing to the factored matrices

    
[out]
Xptr
array of pointers to dense storage of solution

    
[in]
ldX
array of leading dimensions of X

    
[out]
Berrs
array of poiniters to backward errors

    
[in]

  
  

     grid3d contains MPI communicator
Parameters

  

    
[out]
stat
records algorithms statistics such as runtime, memory usage, etc.

    
[out]
info
flags the errors on return



  
  

!!! CHECK SETTING: TO BE SURE TO USE GPU VERSIONS !!!! gpu3dVersion superlu_acc_offload

Here is the call graph for this function:

Here is the caller graph for this function: