Table des matières

Cusp

Cusp is a library for sparse linear algebra and graph computations on CUDA. Cusp provides a flexible, high-level interface for manipulating sparse matrices and solving sparse linear systems. Get Started with Cusp today!

Version installée

Utilisation

Pour utiliser CUSP il suffit de se connecter sur les machines tesla et de charger l'environnement CUDA
$ qlogin -q tesla.q
$ module load gpu/cuda

Exemples

** Calcul le transposé d'une matrice **

transpose.cu
#include <cusp/transpose.h>
#include <cusp/array2d.h>
#include <cusp/print.h>
 
int main(void)
{
    // initialize a 2x3 matrix
    cusp::array2d<float, cusp::host_memory> A(2,3);
    A(0,0) = 10;  A(0,1) = 20;  A(0,2) = 30;
    A(1,0) = 40;  A(1,1) = 50;  A(1,2) = 60;
 
    // print A
    cusp::print(A);
 
    // compute the transpose
    cusp::array2d<float, cusp::host_memory> At;
    cusp::transpose(A, At);
 
    // print A^T
    cusp::print(At);
 
    return 0;
}

** Chargement d'une matrice depuis un fichier (MatrixMarket), résolution du système A.X=b sur GPU en utilisant la méthode GC.**

cg.cu
#include <cusp/hyb_matrix.h>
#include <cusp/io/matrix_market.h>
#include <cusp/krylov/cg.h>
 
int main(void)
{
    // create an empty sparse matrix structure (HYB format)
    cusp::hyb_matrix<int, float, cusp::device_memory> A;
 
    // load a matrix stored in MatrixMarket format
    cusp::io::read_matrix_market_file(A, "5pt_10x10.mtx");
 
    // allocate storage for solution (x) and right hand side (b)
    cusp::array1d<float, cusp::device_memory> x(A.num_rows, 0);
    cusp::array1d<float, cusp::device_memory> b(A.num_rows, 1);
 
    // solve the linear system A * x = b with the Conjugate Gradient method
    cusp::krylov::cg(A, x, b);
 
    return 0;
}

Compilation

Il suffit d'utiliser le compilateur Nvidia nvcc
$ nvcc transpose.cu -o transpose

Liens

http://code.google.com/p/cusp-library/