Skip over navigation

keyword search: ITC

For:

• Students
• Instructors
• Researchers
• Staff
• IT Professionals

Compiling Matlab M-Files on the Linux and Solaris Clusters

This document describes how to use the Matlab compiler to compile and run Matlab M-files as multiple concurrent standalone applications on the Aspen Linux cluster, the Cedar Linux cluster, and the Oak Solaris cluster. This allows users to run the same Matlab program with different inputs (e.g. parameter space studies) effectively in parallel, and without using multiple Matlab network licenses.

This page will be modified to include the Oak Solaris cluster.

Table of Contents

• Matlab Compiler Overview
• Configuring Your Account
• Using the Compiler
• Example M-file for Concurrent Simulations
• PBS Job Command File

Matlab Compiler Overview

The Matlab Compiler takes function M-files as input and creates redistributable, stand-alone applications or software components that are platform specific. The Matlab Compiler can generate

• Stand-alone applications applications that do not require Matlab at run-time; they can run even if Matlab is not installed on the end-user's system.
• C and C++ shared libraries (dynamically linked libraries, or DLLs, on Microsoft Windows). These can be used without Matlab on the end-user's system.
• Excel add-ins; requires Matlab Builder for Excel.
• COM objects; requires Matlab Builder for COM.

The advantages of compiling matlab M-Files as stand-alone executables to run on the Linux clusters are,

• Possible speed-up in program run-time.
• Standalone executable can be run outside Matlab environment.
• No Matlab network licenses required to run program.
• Allows multiple Maltab programs to run concurrently without using multiple network licenses.
• Allows multiple instances of the same program with different inputs to run concurrently, e.g. parameter space studies or Monte Carlo simulations. Effectively provides "embarrassing parallel" processing.

The following is a link to the MathWorks Matlab Compiler Documentation.

Configuring Your Account

1. Edit your startup file (e.g. .variables.ksh) so that your LD_LIBRARY_PATH variable automatically includes the platform specific Matlab libraries needed by the standalone executable. For example, you could add the following statements (verbatim) to your .variables.ksh file.

   # Set LD_LIBRARY_PATH for Matlab standalone executable 
   module add mcc
   

   This module sets the LD_LIBRARY_PATH variable for the Matlab libraries and will auotmatically be updated with each new version of the compiler.

   If you have previously set LD_LIBRARY_PATH using the version specific directory paths to the Matlab libraries, e.g.

   OS=`/uva/bin/cpuarch`
   case $OS in
     Linux ) LD_LIBRARY_PATH=/common/matlab/7.01/bin/glnx86:\
   /common/matlab/7.01/sys/os/glnx86:\
   /common/matlab/7.01/sys/java/jre/glnx86/jre1.4.2/lib/i386/client:\
   /common/matlab/7.01/sys/java/jre/glnx86/jre1.4.2/lib/i386:\
   /common/matlab/7.01/sys/opengl/lib/glnx86:${LD_LIBRARY_PATH} 
   XAPPLRESDIR=/common/matlab/7.01/X11/app-defaults ;;
   esac
   

   you should delete these lines from your startup file.

2. To make sure you are using the system default Matlab compiler options file rather than a local one in your account, run the following command from the Linux command line to remove any local compiler options file from your account.

    rm ~/.matlab/R14/mbuildopts.sh 
   

Using the Compiler

To test the compiler and account configuration copy the example M-file /common/matlab/7.1/extern/examples/compiler/magicsquare.m to your own account, From within Matlab compile the function magicsquare.m as follows:

>> mcc -m -v magicsquare

• The -m option tells the Matlab Compiler (mcc) to generate a C stand-alone application.
• The -v option (verbose) displays the compilation steps throughout the process.
• If your program does not use graphics, you should also include the compiler options '-R -nojvm' in the command line above to disable the Java Virtual Machine.
• Additional M-files called by the main M-file would be appended to the compilation command line.

magicsquare 4 

ans =
    16     2     3    13
     5    11    10     8
     9     7     6    12
     4    14    15     1

Example M-file for Concurrent Simulations

• The standalone executable reads both character and numeric data from a user-defined file.
• The standalone executable reads in and writes out data to simulation-specific named files.

function matlab_sim()

% Read in program initialization parameters from text file 
% e.g. see sim_input1 below 

% First read in a character string to be used to identify the simulation number
% e.g. the character 1.
sim_num=input('simulation identifier > ','s')

% Example of reading in numeric vector  used for initialization for simulation
vector_input=eval(input('vector input > ','s'))

% Example of loading data from .mat file with filename based on simulation number,
% e.g. file_in1.mat
% This assumes the .mat file already exists containing some data.
load(['file_in',sim_num,'.mat'])

% Other Matlab code can be included in the body of this function M-file.

% Create some matrix a
a=rand(5,5);

% Save variable a to .mat file with filename based on simulation number
% e.g. file_out1.mat
save(['file_out',sim_num], 'a')

% Alternatively, save variable a to ascii file with filename based on simulation number
% e.g. junk1
fid = fopen(['junk',sim_num],'w');
fprintf(fid,'%12.8f %12.8f %12.8f %12.8f %12.8f \n',a);
fclose(fid);

matlab_sim < sim_input1

The input file for matlab_sim is sim_input1 and contains the following data:

1
[3 6 9]

This way you can run different concurrent simulations by using different versions of the text file sim_input1 (e.g. sim_input2, sim_input3, etc.) without having to recompile your code.

PBS Job Command File

The following is an example of a PBS command file used to submit the Matlab stand-alone exectuable matlab_sim (after compilation) for the specific simulation defined by the input file sim_input1 to the Linux clusters. Further information on PBS command files can be found in the Aspen Cluster Tutorial

#!/bin/sh
#PBS -l nodes=1:ppn=1
#PBS -l walltime=12:00:00
#PBS -o output_filename
#PBS -j oe
#PBS -m bea
#PBS -M userid@virginia.edu

cd $PBS_O_WORKDIR
./matlab_sim < sim_input1