MATLAB
Introduction
MATLAB is a powerful scripting language and computational environment. It is designed for numerical computing, visualization and high-level programming and simulations. MATLAB also has parallel processing capabilities.
MATLAB is installed on all HPC compute and login nodes, and on all Spear nodes.
Limited Licenses
We maintain a limited number of licenses for MATLAB use at the RCC. Sometimes, all available licenses are in-use by other users and you must wait to checkout a license. To see how many licenses are available, you can run the following command from inside MATLAB:
license('checkout','MATLAB')
Non-interactive jobs submitted using the scripts shown in this page will check for available MATLAB licenses before running. However, if you want to use large number of MATLAB jobs simultaneously, use MATLAB Compiler as described below to create executables from your code. This will compile your MATLAB code to C code and avoid all license restrictions.
MATLAB Profiles
In order to submit MATLAB jobs to the HPC, you must have a MATLAB profile. This only has to be done once per user per MATLAB version. This means when we occasionally upgrade MATLAB, you may need to recreate your profile.
Creating a MATLAB Profile on the HPC
Copy the profile file to your MATLAB working directory using the following command (assuming your are already in your working directory or home directory):
cp /panfs/storage.local/opt/userfiles/hpc.settings .
Run MATLAB on the HPC (login with -Y to enable graphical applications):
ssh -Y YOUR_ACCOUNT@hpc-login.rcc.fsu.edu
$ matlab
From the Home tab in MATLAB, find the Parallel dropdown. From this dropdown, choose Manage Cluster Profiles.
Click Import and browse to the location where you downloaded "hpc.settings" file. Click "Open"
Close the Cluster Profile Manager window.
Alternatively, you can also import the profile from within the command-line interface of MATLAB:
hpc = parallel.importProfile('/panfs/storage.local/opt/userfiles/hpc.settings');
Creating a MATLAB Profile on your own copy of MATLAB
If you already have a copy of MATLAB on your own computer, you can use that to submit MATLAB jobs to the HPC. If you are off campus, be sure that you are connected to the VPN.
The profile for this is slightly different than the above procedure, and it can be created on your own laptop or desktop as follows.
From the Home tab in MATLAB, find the Parallel dropdown. From this dropdown, choose Manage Cluster Profiles.
Click Add and select Custom >> Generic from the menu.
Right click on the "GenericProfile1" on the left pane and rename it to "HPC". Your profile is setup, and you can submit jobs following the procedure below in the Job Submission from your own copy of MATLAB section below.
Interactively running MATLAB
You can work with MATLAB interactively on our servers, similar to how you would on your own workstation by using the Spear cluster. This is useful for running short jobs or testing/debugging production runs. You can run MATLAB on Spear by connecting to Spear, opening a terminal, and then executing the MATLAB program:
[user@spear-login.rcc.fsu.edu] $ module load matlab
[user@spear-login.rcc.fsu.edu] $ matlab
You can use the MATLAB Parallel Computing Toolbox (PCT) to utilize more than one core. To use any of the parallel features of MATLAB (such as parfor), there are to modes:
- pmode is an interactive mode where you see individual workers (labs) in a GUI
- mathlabpool is the mode where the labs run in the background.
The maximum number of workers that can be used in either mode is limited to 12.
Using pmode interactively
Below is an example of how to invoke interactive pmode with four workers:
pmode start local 4
This will open a Parallel Command Window (PCW). The workers then receive commands entered in PCW (at the P>> prompt), process them, and send the command output back to the PCW. You can transfer variables between the MATLAB client and the workers. For example, to copy the variable x in Lab 2 to xc on the client, use:
pmode lab2client x 2 xc
Similiarly, to copy the variable xc on the client to the variable x on Lab 2, use:
pmode client2lab xc 2 x
You can perform plotting and all other operations from inside PCW.
You can distribute values among workers as well. For example, to distribute the array x among workers, use:
codistributed(x,'convert')
Use numlabs
, labindex
, labSend
, labReceive
, labProbe
, labBroadcast
, labBarrier
functions similar to MPI commands for parallelizing. Please refer to MATLAB manual for a full discussion of commands. Entering the following command in the PCW will end the session and release the licenses for other users:
pmode quit
Using matlabpool interactively
In the following example, you can invoke mathlabpool
mode with 4 workers using:
matlabpool local 4
You can close the pool as follows:
matlabpool close
It is also possible to invoke mathlabpool using the batch command. For example:
batch('test1','matlabpool', 2);
In this example, test1.m is a MATLAB script, and we want to run it with 2 workers. The program runs in the background and will have to direct the output to be written into files for longer running jobs. The advantage of this approach is that you can create sets of jobs using same function with different input parameters. For example:
batch('matlabpool', 'test1', number of output arguments, {x1,...,xn});
...where x1, ..., xn are different function inputs.
Interactive parallel computing using parpool
The matlabpool
utility will be replaced by the parpool
utility in future versions of MATLAB. The syntax for parpool
is
parpool
parpool(poolsize)
parpool('profile',poolsize)
parpool('cluster',poolsize)
ph = parpool(...)
...where poosize
, profile
, and cluster
are respectively the size of the MATLAB pool of workers and the profile or the cluster you created. The last line creates a handle ph
for the pool.
parpool
enables the full functionality of the parallel language features (parfor
and spmd
) in MATLAB by creating a special job on a pool of workers, and connecting the MATLAB client to the parallel pool.
The following example creates a pool of 4 workers, and runs a parfor-loop
using this pool:
>>parpool(4)
>> parfor i = 1:10
feature getpid;
disp(ans)
end
1172
1172
1171
1171
1169
1169
1170
1172
1171
1169
>> delete(gcp)
The following example creates a pool of 4 workers, and runs a simple spmd
code block using this pool:
>> ph = parpool('local',4) % ph is the handle of the pool
>> spmd
>> a = labindex
>> b = a.^2
>> end
Lab 1:
a = 1
b = 1
Lab 2:
a = 2
b = 4
Lab 3:
a = 3
b = 9
Lab 4:
a = 4
b = 16
>> delete(ph)
Note. You cannot simultaenously run more than one interactive parpool
session. You must to delete your current parpool session before starting a new one. To delete the current session, use:
delete(gcp)
delete(ph)
...where gcp
utility returns the current pool, and ph
is the handle of the pool.
Non-interactive job submission
The following examples demonstrate how to submit MATLAB jobs to the HPC. You should already be familiar with how to connect and submit jobs in order to submit MATLAB jobs.
Single core jobs
The following example is a sample submit script (test1.sh) to run the MATLAB program test1.m that uses a single core.
Note: test1.m should be a function and not a script.
#!/bin/bash
#SBATCH -N 1
#SBATCH --ntasks-per-node=1
#SBATCH -p genacc_q
#SBATCH -t 01:00:00 #Change the walltime as necessary
module load matlab
matlab -nosplash -nojvm -nodesktop -r "test1 exit"
Then, you can submit your job to Slurm normally:
sbatch test1.sh
Note: The Parallel Computing Toolkit (PCT) cannot be used within your function; you cannot use parfor or any other command that utilizes more than one core.
Multiple Core Jobs
You can submit MATLAB jobs to run on multiple cores in a single processor by changing the ppn value in the MATLAB submit script. The maximum number of cores available for any single job is currently eight.
#!/bin/bash
#SBATCH -N 1
#SBATCH --ntasks-per-node=4 # Use 4 cores in a single processor (change as needed; up to 8)
#SBATCH -p genacc_q
#SBATCH -t 01:00:00 # Change the wall time as necessary
module load matlab_dcs
matlab -nosplash -nodesktop <<EOF
matlabpool open local 4 # This should match the value in line 3
test2 #This is your MATLAB function
matlabpool close
exit
EOF
Multiple node jobs
The above examples demonstrate how to submit single-processor jobs. You can use the MATLAB Distributed Computing Engine (MDCE) to submit multi-processor, multi-node jobs and take advantage of the massive parallelization of the HPC.
Before using MDCE, you must have a MATLAB profile (see sction above), and you must load the matlab_dcs module:
module load matlab_dcs
matlab
You can use the following template to submit jobs from MATLAB command line on any HPC login node. Make sure that the matlab_work directory exists, or change line 3 to point to the foldirectoryer where you store your MATLAB code.
Also, change $USER in line 3 to your username, and change the name of the HPC partition in line 4 to whichever you wish to submit to.
cluster = parcluster('hpc');
% REPLACE $USER in the following line with your username and ensure the "matlab_work"
% directory exists
set(cluster, 'JobStorageLocation', '/panfs/storage.local/genacc/home/$USER/matlab_work');
% Change the partition name if you wish to submit to if desired
set(cluster, 'SubmitArguments', '-p genacc_q');
% Change the number of workers below as desired. This example shows 4 tasks on 2 nodes.
% Change the wall time parameter. This example shows a 10 minute job.
set(cluster, 'ResourceTemplate', '-n 4 -N 2 -t 00:10:00');
set(cluster, 'NumWorkers', 4);
% Note that MATLAB needs n+1 workers assigned to it, because there is one overhead process
% For example, if you assign 4 workers, only 3 will do actual work
% Create an independent job
j = createJob(cluster, 'Profile', 'hpc');
% Create tasks:
% This "test3" FUNCTION accepts one input ('100' in this example) and generates one output value
createTask(j, @test3, 1, {100});
% This example accepts two inputs
%createTask(j, @test3, 1, {{100},{200}});
submit(j)
% Collect output. You need to wait until the job completes!
% If you use the wait(j) command, you will not get the MATLAB prompt back until all jobs are finished
% Fetch output, if you are writing them to STDOUT.
% It is probably better if your MATLAB script writes directly to an output file
taskoutput = fetchOutputs(j);
% VERY IMPORTANT! delete job object
delete(j)
You can create a communicating job by replacing the createJob
command with the following:
j = createCommunicatingJob(cluster, 'Profile', 'hpc');
Using the MATLAB Compiler
If you need to use a large number of simultaneous MATLAB workers, we advise that you compile your code into an executable. This allows you to write and test your code in MATLAB, and compile it to C when you are ready to run a production job. One major advantage of compiling your MATLAB code is that your job will not be restricted by license limitations.
You can use the MATLAB compiler, mcc
, on Spear or any of the HPC login nodes in order to create a binary executable of the code. Be sure to compile in whichever environment (HPC or Spear) you intend to run the code.
To compile the non-parallel code test1.m, use the MATLAB command:
mcc -R -nodisplay -R -nojvm -R -nosplash -R -singleCompThread -m test1.m
Note: This works only with the matlab module, not the matlab_dcs module.
The above command will create the script run_test1.sh and the executable test1. A brief description of these compiler flags follows.
-R: Specifies runtime options, and must be used with the other runtime flags (nodisplay, nosplash, etc)
-R -nodisplay: Any functions that provide a display will be disabled
-R -nojvm: Disable the Java Virtual Machine
-R -nosplash: Starts MATLAB, but does not display the splash screen
-R -singleCompThread: Runs only a single thread in the runtime environment
-m: Generates a C binary (-p would generate a C++ binary)
After successful completion, mcc
creates the following files:
- a binary file,
- a script to load necessary environment variables and run the binary,
- a readme.txt, and,
- a log file.
The binary file can be run via the generated script with the following command:
<full-path-to-script>run_test1.sh /opt/matlab/current <input-arguments>
You can place this command in your Slurm submit script in order to run it as part of an HPC job.
Note that input arguments will be interpreted as string values, so any code that utilizes these arguments must convert these strings to the correct data type.
One concern with this method of compiling and running a MATLAB program is that the binaries generated will contain all of the toolboxes available in the user's MATLAB environment, resulting in large binary files. To avoid this, use the -N
compiler flag. This will remove all but essential toolboxes, and other tools or .m
files required can be attached with the -a
option. The recommended syntax for generating serial binaries is:
mcc -N -v -R -nodisplay -R -nojvm -R -nosplash -R -singleCompThread -m test1.m
In this example, the -v
option enables verbose mode. To run a MATLAB program in parallel using the Parallel Computing Toolbox, add the -p distcomp
option to the command line arguments,
mcc -N -v -p distcomp -m test1.m
To run this binary, you must provide a parallel profile (see Creating a Profile above):
<full-path-to-script>run_test1.sh /opt/matlab/current -mcruserdata ParallelProfile:/opt/matlab/r2013b/toolbox/distcomp/parallel.settings
In this example, the default local
profile is used. This can be used in any SLURM submit script without any modification, and no MATLAB licenses will be used when the job runs.
The generation of binaries is the same for MATLAB Distributed Computing Engine jobs, but a different SLURM-aware profile needs to be provided. You should always used the supplied HPC
profile for these types of jobs. See the section above on submitting paraellel MATLAB jobs to see what lines should be added to a MATLAB script when using the Distributed Computing Engine. The submit command for a distributed job is:
<full-path-to-script>run_test1.sh /opt/matlab/current -mcruserdata ParallelProfile:/panfs/storage.local/opt/userfiles/hpc.settings
Using MATLAB with GPU Processors
MATLAB is capable of using GPUs to accelerate calculations. Most built-in functions have alternative GPU versions. In order to take advantage of GPUs, you will need to submit your jobs to the HPC gpu_q partition. This will schedule your jobs to run on our GPU-enabled compute nodes.
You can try the following example, which is a mandelbrot program, f_mandelbrot:
function [mbset, t] = gpu_mandelbrot(niter, steps, xmin, xmax, ymin, ymax)
t0 = tic();
x = gpuArray.linspace(xmin, xmax, steps);
y = gpuArray.linspace(ymin, ymax, steps);
[xGrid,yGrid] = meshgrid(x, y);
c = xGrid + 1i * yGrid;
z = zeros(size(c));
mbset = zeros(size(c));
for ii = 1:niter
z = z.*z + c;
mbset(abs(z) > 2 & mbset == 0) = niter - ii;
end
t = toc(t0);
Run the program and display the results with the following commands:
[mandelSet, time] = gpu_mandelbrot(3600,100,-2,1,-1.5,1.5)
surface(mandelSet)
The arrays x
and y
are generated on the GPU, utilizing its massively parallel architecture, which also handles the remainder of the computations that involve these arrays.
Submitting HPC jobs from your own copy of MATLAB
This section describes how to submit MATLAB jobs from your own workstation (desktop/laptop) to the HPC cluster. Use VPN if you are off campus.
The following figure illustrates the scheme for remote job submissions:
To submit a MATLAB job to the HPC cluster from your computer, a generic scheduler interface must be used. You must create a generic cluster object in MATLAB before you can submit jobs:
Function | Description |
---|---|
CancelJobFcn | Function to run when cancelling job |
CancelTaskFcn | Function to run when cancelling task |
CommunicatingSubmitFcn | Function to run when submitting communicating job |
DeleteJobFcn | Function to run when deleting job |
DeleteTaskFcn | Function to run when deleting task |
GetJobStateFcn | Function to run when querying job state |
IndependentSubmitFcn | Function to run when submitting independent job |
We have provided these functions in the tarball matlab.tar
located at:
/panfs/storage.local/opt/userfiles/matlab.tar
These functions should be copied to:
[LOCAL_MATLAB_ROOT_DIRECTORY]/toolbox/local/
Substitute the path to MATLAB on your workstation for [LOCAL_MATLAB_ROOT_DIRECTORY].
You can use the following commands to download and extract the files (assuming you are in your MATLAB root directory) on your workstation:
$ scp you@hpc-login.rcc.fsu.edu:/panfs/storage.local/opt/userfiles/matlab.tar .
$ tar -xvf matlab.tar
matlab/
matlab/communicatingSubmitFcn.m
matlab/independentSubmitFcn.m
matlab/getSubmitString.m
matlab/communicatingJobWrapper.sh
matlab/createSubmitScript.m
matlab/extractJobId.m
matlab/deleteJobFcn.m
matlab/getRemoteConnection.m
matlab/getJobStateFcn.m
matlab/independentJobWrapper.sh
matlab/getCluster.m
Next, you must setup PKI authentication (ssh-key-based instead of password) to the HPC. Refer to our "Using SSH Keys" documentation for how to do this. Make sure that you do NOT use a passphrase when setting up your keypair.
Now, move all the files except getCluster.m
to the toolbox/local directory on your machine. For example, if you installed MATLAB to /usr/local directory, this can be done with:
mv matlab/!(getCluster.m) /usr/local/MATLAB/R2013b/toolbox/local/
Make sure the path is correct before you copy and that you have administrative privileges on your computer.
Next, update the `getCluster.m` file. matlab.tar
provides several functions you need to create a generic
cluster object. The function getCluster()
use these functions to configure a cluster object for you.
Here is the content of the getCluster.m:
function [ cluster ] = getCluster(ppn, queue, rtime, LocalDataLocation, RemoteDataLocation)
%Find the path to id_rsa key file in YOUR system and update the following line
username = 'YOUR HPC USER NAME'
keyfile = '/home/USER/.ssh/id_rsa'; %Your actual path may be DIFFERENT!
%Do not change anything below this line
if (strcmp(username, 'YOUR HPC USER NAME') == 1)
disp('You need to put your RCC user name in line 4!')
return
end
if (exist(keyfile, 'file') == 0)
disp('Key file path does not exist. Did you configure password-less login to HPC?');
return
end
ClusterMatlabRoot = '/opt/matlab/current';
clusterHost='submit.hpc.fsu.edu';
cluster = parcluster('hpc');
set(cluster,'HasSharedFilesystem',false);
set(cluster,'JobStorageLocation',LocalDataLocation);
set(cluster,'OperatingSystem','unix');
set(cluster,'ClusterMatlabRoot',ClusterMatlabRoot);
set(cluster,'IndependentSubmitFcn',{@independentSubmitFcn,clusterHost, ...
RemoteDataLocation,username,keyfile,rtime,queue});
set(cluster,'CommunicatingSubmitFcn',{@communicatingSubmitFcn,clusterHost, ...
RemoteDataLocation,username,keyfile,rtime,queue,ppn});
set(cluster,'GetJobStateFcn',{@getJobStateFcn,username,keyfile});
set(cluster,'DeleteJobFcn',{@deleteJobFcn,username,keyfile});
The five input arguments of the function getCluster
are
Argument | Description |
---|---|
ppn | processor/core per node |
queue | queue you want to submit job to (e.g., backfill, genacc_q) |
rtime | wall time |
LocalDataLocation | directory to store job data on your workstation |
RemoteDataLocation | directory to store job data on your HPC disk space |
The first time you download, edit the getCluster.m file to include the following:
- your RCC user name in line 4, and
- the correct path to your id_rsa key file in line 5 (usually ~/.ssh/id_rsa)
Before you call this function, create a separate folder (to be used as "RemoteDataLocation" in the getCluster.m
script) in your HPC disk space to store runtime MATLAB files, for example,
[user@hpc-login.rcc.fsu.edu] $ mkdir -p $HOME/matlab/work
Also create a folder in your workstation to be used as the LocalDataLocation
. Clean these folders regularly after finishing jobs; they tend to fill up.
Submitting Jobs
The following lines can be used as a template to create a generic
cluster object which you will use to submit jobs to HPC from your local copy of MATLAB:
processors = 4; % Number of processors used. MUST BE LESS THAN OR EQUAL TO 32
ppn = 4; % Number of cores used per processor
queue = 'genacc_q'; % Replace this with your choice of partition
time = '01:00:00'; % Run time
LocalDataLocation = ''; % Full path to the MATLAB job folder on your workstation (not the HPC)
% Full path to a MATLAB scratch folder on HPC (replace USER with your RCC username)
RemoteDataLocation = '/panfs/storage.local/genacc/home/USER/matlab/work';
cluster = getCluster(ppn, queue, time, LocalDataLocation, RemoteDataLocation);
The following is an example to create acommunicating job
for thecluster
:
j1 = createCommunicatingJob(cluster); % This example creates a communicating job (eg:parfor)
j1.AttachedFiles = {'testparfor2.m'}; % Send all scripts and data files needed for the job
set(j1, 'NumWorkersRange', [1 processors]); % Number of processors
set(j1, 'Name', 'Test'); % Give a name for your job
t1 = createTask(j1, @testparfor2, 1, {processors-1});
submit(j1);
%wait(j1); % MATLAB will wait for the completion of the job
%o=j1.fetchOutputs; % Collect outputs after job is done
Note. Only use the last two lines in this example for testing small jobs. Production jobs may wait in the HPC queue for a long time and will cause your copy of MATLAB to wait until the job has completed before being usable again. For production jobs, you should write data directly to the filesystem from your MATLAB script instead of collecting it when the script is complete.
The following is an example to create an independent job
for the cluster
:
j2 = createJob(cluster); % create an independent job
t2 = createTask(j2, @rand, 1, {{10,10},{10,10},{10,10},{10,10}}); % create an array of 4 tasks
submit(j2)
wait(j2)
o2 = fetchOutputs(j2) % fetch the results
o2{1:4} % display the results
Note. A communicating
job contains only one task. However, this task can run on multiple workers. Also, the task can contain parfor-loop
or spmd
code block to improve the performance. Conversely, an independent
job can contain multiple tasks. These tasks do not communicate with each other and each task runs on a single worker.
MATLAB Workshops
We regularly offer MATLAB workshops. Keep an eye on our event schedule or send us a message to find out when the next one will be. You can also check out our slides and materials from past workshops.