Skip to content

FacundoGFlores/NSL

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 

Repository files navigation

Description

This program was presented as a final project of the subject called: "Seminar" in April 2013.

The project implements many iterative methods for solving sys- tem of linear equations.

Usage

I implemented a generic code for showing the use of the different created libraries. I use a generic matrix for illustrative purpose. You can take any of the libraries for solving a specific kind of matrix.

Example:

linearMatrix A(N,N);
linearMatrix B(N,1);
linearMatrix X(N,1);

//Call a rutine to prepare matrices
//Prepare(A,B,X);
//Remember you can use the following method from "linearMatrix.hpp"
//void Read_File_Matrix(const char *Path);
//Read documentation for more info about matrix-file structure

Parallel P(A,B,X, error, iter); //Object with parallel methods
P.With_Jacobi(); //P.With_"METHOD_NAME"

linearMatrix SOLUTION(N,1);
SOLUTION = P.Get_Solution();

//example: SOLUTION.Write_File_Matrix("Solution.dat");

Repo Installation

I used the Nsight Eclipse IDE on Linux. I tested the project on the 5 and 5.5 editions.

  • Copy the NSL repo on your own cuda-workspace
  • Go to File -> Import -> Existing Projects into Workspace
  • Select the NSL repo
  • Click Finish

Compiling

To configure the compiler please go:

  • Project -> Properties -> Build -> Settings

Then edit into the NVCC Compiler the "All options" section:

-l/home/YOUR_HOME/cuda-workspace/NSL/include -G -g -O0

Errata

In the perfomance timing analysis I put the starting clocks after the cublas structures initializations, so for better performance timing analysis put the clocks before the cublas initializations.

About

Solving System of Linear Equations with CUDA and cuBLAS

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published