1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
|
/*
-- MAGMA (version 2.5.4) --
Univ. of Tennessee, Knoxville
Univ. of California, Berkeley
Univ. of Colorado, Denver
@date October 2020
@precisions normal d -> s
@author Hartwig Anzt
*/
// includes, system
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <math.h>
// includes, project
#include "magma_v2.h"
#include "magmasparse.h"
#include "testings.h"
/* ////////////////////////////////////////////////////////////////////////////
-- This is an example how magma can be integrated into another software.
*/
int main ()
{
// The software does e.g. discretization of a PDE,
// ends up with a sparse linear system in CSR format and a RHS.
// Let's assume this system is a diagonal system of size m.
int m=200, n=1;
int *col, *row;
double *val, *rhs, *sol;
row = (int*) calloc (m+1, sizeof(int));
col = (int*) calloc (m, sizeof(int));
val = (double*) calloc (m, sizeof(double));
rhs = (double*) calloc (m, sizeof(double));
sol = (double*) calloc (m, sizeof(double));
for( int i=0; i<m; i++ ){
col[i] = i;
row[i] = i;
val[i] = 55.0;
rhs[i] = 3.0;
sol[i] = 0.0;
}
row[m] = m;
// We now initialize MAGMA and create some LA structures.
magma_init();
magma_dopts opts;
magma_queue_t queue;
magma_queue_create( 0, &queue );
magma_d_matrix A={Magma_CSR}, dA={Magma_CSR};
magma_d_matrix b={Magma_CSR}, db={Magma_CSR};
magma_d_matrix x={Magma_CSR}, dx={Magma_CSR};
// We now pass the system to MAGMA.
magma_dcsrset( m, m, row, col, val, &A, queue );
magma_dvset( m, 1, rhs, &b, queue );
magma_dvset( m, 1, sol, &x, queue );
// We now choose a solver, preconditioner etc - see documentation for options.
opts.solver_par.solver = Magma_PIDRMERGE;
opts.solver_par.restart = 8;
opts.solver_par.maxiter = 1000;
opts.solver_par.rtol = 1e-10;
opts.solver_par.maxiter = 1000;
opts.precond_par.solver = Magma_ILU;
opts.precond_par.levels = 0;
opts.precond_par.trisolver = Magma_CUSOLVE;
// We initialize the solver.
magma_dsolverinfo_init( &opts.solver_par, &opts.precond_par, queue );
// We copy the system to the device (optional, only necessary if using the GPU)
magma_dmtransfer( A, &dA, Magma_CPU, Magma_DEV, queue );
magma_dmtransfer( b, &db, Magma_CPU, Magma_DEV, queue );
magma_dmtransfer( x, &dx, Magma_CPU, Magma_DEV, queue );
// We now generate the preconditioner.
magma_d_precondsetup( dA, db, &opts.solver_par, &opts.precond_par, queue );
// In case we only wanted to generate a preconditioner, we are done.
// The preconditioner in the opts.precond_par structure - in this case an ILU.
// The lower ILU(0) factor is in opts.precond_par.L and
// the upper ILU(0) factor is in opts.precond_par.U (in this case on the device).
// If we want to solve the problem, we run:
magma_d_solver( dA, db, &dx, &opts, queue );
// Then we copy the solution back to the host...
magma_dmfree( &x, queue );
magma_dmtransfer( dx, &x, Magma_DEV, Magma_CPU, queue );
// and back to the application code
magma_dvget( x, &m, &n, &sol, queue );
// We free the allocated memory...
magma_dmfree( &dx, queue );
magma_dmfree( &db, queue );
magma_dmfree( &dA, queue );
// and finalize MAGMA.
magma_queue_destroy( queue );
magma_finalize();
// From here on, the application code may continue with the solution in sol...
for(int i=0; i<20; i++){
printf("%.4f\n", sol[i]);
}
}
|
