MATLAB and Simulink
MathWorks Inc. | R2014b Dec 2014Linux/Unix, Ubuntu 12.04 - 64-bit Amazon Machine Image (AMI)
External reviews
External reviews are not included in the AWS star rating for the product.
Worked because we followed the directions
We tested MATLAB on EC successfully, using our lab's Standalone Named User license.
The directions from Mathworks "MATLAB on EC2 Instructions" are long and detailed, but worked, except for one or two points. We needed to restart nxserver on the server before we were able to connect with the client. Do that with this command via SSH:
$ sudo /usr/lib/nx/nxserver --restart
After that, we connected using the OpenNX client (Mac OS X). Another problem, the .pem key file imported by the OpenNX client had problems with spaces/extra carriage returns introduced by cut-and-paste (you can avoid that problem by generating the key pair on your local machine and uploading the public key to AWS).
Matlab cannot be launched from SSH, which is a shame (it seems like the GUI is needed to do the authentication part). Fixing this would be a very welcome improvement.
Our license did not include the parallel processing toolbox, which means we could not do real work. We'd have to upgrade the license to do anything useful.
On EC, we used c4.xlarge and t2.medium instances. We tried the 36-core c4.8xlarge, but unfortunately the instance never launched. MATLAB on EC2 cannot be used at the AWS's free tier.
- Leave a Comment |
- 1 comment |
- Mark review as helpful
Does the job, and is super fast
Every time you start Matlab, you have to sign in with your Mathworks username and password, which is kind of a drag; it also means you need to run Matlab using the GUI instead of from the terminal with the -nodisplay flag (since, if you use -nodisplay, the prompt to enter your username and password doesn't appear and Matlab doesn't start). So minus one star for that.
Otherwise, when using the parallel computing toolbox, things can be blazing fast with a beefy EC2 instance, which is totally great. With matlab R2014a, they removed the 12-worker limit, so you can have as many workers in your parallel pool as you have (virtual) CPUs in the EC2 instance. Computing with 32 cores (via a c3.8xlarge instance) is simple and super quick.
In response to the other reviewers concerns that on-demand pricing isn't available, on-demand usage of the parallel computing toolbox on EC2 is available at http://www.mathworks.com/discovery/matlab-ec2.html. But you need to already have a Matlab and parallel computing toolbox license.