Eigen is a free and open source, relatively fast, very versatile linear algebra library written completely in C++ templates. It is very fast and in some cases produces code as fast as commercial implementations like Intel MKL, AMD ACML etc. The beauty of the Eigen is, it’s completely written in C++ templates and it’s very easy to plug it in to a project compiled with any supported compiler.

Here I describe the process of how to install Eigen with Microsoft Visual Studio 2010 Express and 64bit compilers from Microsoft Windows SDK 7.1 supporting OpenMP and vectorization with SSE2.  All these tools are free.

Install Microsoft Tools

Install Visual C++ 2010 Express Edition &
Install Microsoft Windows SDK for Windows Version 7.1.


When WindowsSDK is installed,
Make sure to include Visual C++ Compilers as seen in the above picture

Install Eigen Library

Just unzip the latest version of the Eigen template library to the hard disk.

Creating New Project

Start Visual c++ 2010 Express and create a new
Visual C++ Empty Project. Give a name for the new project.

Add a new c++ source file to the project

Add a new c++ source file to the project and copy-paste the code below. Visual C++ editor will highlight missing include files and datatypes in the source code. This is expected because Visual C++ environment still doesn’t know where the Eigen libraries, and this can be fixed very easily.

#include <iostream>
#include <ctime>
#include <Eigen/Eigen>

using namespace std;
using namespace Eigen;
int main()
{

MatrixXd A = MatrixXd::Random(5000,5000);
MatrixXd B = MatrixXd::Random(5000,2);

//cout << "Here is the invertible matrix A:" << endl << A << endl;
//cout << "Here is the matrix B:" << endl << B << endl;

clock_t clock_start=clock();

MatrixXd X = A.lu().solve(B);

double time_c=(double)(clock()- clock_start)/CLOCKS_PER_SEC;

//cout << "Here is the (unique) solution X to the equation AX=B:" << endl << X << endl;

cout << "Relative error: " << (A*X-B).norm() / B.norm() << endl;

cout<< "Elapsed Time is: "<< time_c<< "s" <<endl;

cin.get();
}

Create 64bit solution platforms

Create 64bit "Solution platforms" for both
"Debug" and "Release" "Solution Configurations".

Set Project Properties for ‘Release’ Configuration and ‘x64’ Platform

Platform Toolset – Windows7.1SDK

Set Platform Toolset to Windows7.1SDK

Add Eigen Library to Include Path

Add Eigen library location to VC++ Include path.
Once this step is done, VC++ editor should no longer
highlight error in the sample code.

Multi processor compilation

Turn on Multiprocessor Compilation if you have a multicore processor.

Compiler Optimization

Make sure compiler optimization is set to 'Maximize Speed'

Enable Enhance Instruction Set Support

Enable SSE2 support. Once enabled Eigen will
Make use of SSE2 to Vectorize it's code generation

Turn on OpenMP support

Eigen uses OpenMP when possible. Eigen uses OpenMp specialy
during crucial Matrix-Matrix product calculations.

Build and Run the Project

Build and run the project. The given example code solves a system linear equation with 5000 equations. As you can see from above picture all processor core are occupied during calculation. You can turn on and off OpenMP and SSE2 support and test how fast the code completes when these features are turned on and off.