- GPU Computing. You can't have a MATLAB without CPU Computing so obviously both is better.
- Mostly using the GPU involves the same code, but you convert inputs to the gpuArray datatype. So, if we've done our job right, the transition is not hard.
hardware requierments for MATLAB
2 views (last 30 days)
im trying to build MATLAB enviorment for our univercity and im looking for tips and guidence on what hardware we should use?
i am from IT department of univercity and i have never worked with matlab so i looked to your website and i was not satesfied with information i found so let me ask it here
1- wich one is better ? CPU computing or GPU computing ?
2- can GPU computing proccess all of the MATLAB applications ?
3- does GPU computing have diffrent set of precedures to start computing then CPU computing for students ? i mean should we train them to be able to use GPU computing ?
4- our application for matlab is to proccess Maps and satellite images, regarding our application what hardware do you suggest ?
5- how much SSD does MAtLAB need to run and proccess the task?
6- can users have multiple connections to matlab and do simentusly diffrent tasks without crashing the system ?
7- is it possible to run MATLAB from Virtual Machine ?
8- if MATLAB have been RUN from VM does it have negative performance?
9- do MATLAB have SQL or none SQL base database? if does please clarifie what DB and what performance/IO does MATLAB need?
10- how much RAM does MATLAB need?
Joss Knight on 4 Dec 2019
Jason Ross on 6 Dec 2019
Edited: Jason Ross on 6 Dec 2019
4. The Systems Requrements page explains minimum configurations for Windows, Linux, or Mac hardware. MATLAB runs on any of the specified operating systems. System configuration for running specific applications will be somewhat dictated by the size of the data (for stoarge and RAM), number of cores required, and of course, budget.
5. MATLAB itself varies in installation size, up to 31 GB if you install everything. Typically it's much smaller, about 5-8 GB. Sizing an SSD would also require space for the operating system, other applications, temp space, plus whatever you need to store data and outputs (which are entirely dependent on your workflow)
6. MATLAB runs as a desktop application, so if people are working on independent workstations each application runs separately like any other desktop application. If you are using a shared server, each user logged into the server will have an independent copy running so it might be possible for the users to overwhelm the server, but people will set quotas for the user accounts to prevent this sort of thing. If you have a cluster, we do have products that allow you you to run MATLAB on it, but the cluster scheduler is responsible for dealing with contention.
7. Yes. MATLAB runs as an application on the OS, so it will run inside an virtualized OS.
8. Virtual machines acting through a hypervisor will will have some level of overhead vs. a bare metal install. So MATLAB running on a VM is going to likely be slower than a comprable bare metal OS installation because it (like any other application) will also incur some overhead. The acceptability of this overhead is entirely up to you and your users. Virtualized environments provide a lot of benefits for organizations, and for a lot of people those benefits are worth it.
9. There are no database requirements to run MATLAB. We have means of connecting to databases, but to run the MATLAB application you don't need any database installed.
10. This question is impossible to answer -- it's going to be dependent on what kind of calcualations you are actually doing. For small examples and basic work you can use a very minimal configuration as described on the system requirements page. If you are working with tens, hundreds, or more gigabytes of data you will need more computing resources.