MATLAB Home on Apple Silicon: GPU support?

I'm well aware that I should ask Sales about this but I can't contact them due to the attack. I enter my email on the contact form (same as to login here) to start the process and can't go any further.

I am thinking of resurrecting my MATLAB Home license in my retirement, maybe with Simulink and a couple of toolboxes as well. When I had it before I had an Intel Macbook and now I'm running an M3 Macbook Pro. I used full-up installations at my former employer and found I could learn a lot of new topics through the Toolbox documentation and examples.

I see that MATLAB now has native support for M-series processors and that Intel support will go away soon. What's not completely clear is what I'll have to give up. There's a list of unavailable packages which is a start. https://www.mathworks.com/support/requirements/apple-silicon.html

I would like to explore GPU processing to educate myself on what's possible but can't determine if I can use the GPU on the Mac. There are older Answers threads that say No but I think those were before the native Apple Silicon release.

Also interested in what else I might not be able to do that may not be in the current exclusion lists. The Home license is still an investment and I'd like to avoid any hidden gotchas.

Answer 1

[Damian Pietrus]

Hello Ravi,

At this point in time, MATLAB does not support GPU computing in macOS on either Intel or Apple Silicon platforms. This is tied to the fact that GPU support is limited to NVIDIA® GPU architectures with compute capability 5.0 to 9.x (as of R2025a). Since macOS does not support NVIDIA cards, you unfortunately won't be able to utilize GPU processing.

As for other aspects of the Home License, you can view more information here. Copy/pasting some points below:

What is the difference between home and professional version of MATLAB and Simulink?

MATLAB Home offers you the full capabilities of MATLAB. However, certain add-on products are not available for purchase.

In MATLAB, the Command Window banner and title bar indicate home use.

In Simulink:

• Models (including the blocks in referenced models) are limited to 1000 nonvirtual blocks.
• Accelerator and Rapid Accelerator simulation modes are not available.
• Model blocks can be simulated in Normal mode only

[Ravi Narasimhan]

That's unfortunate but thanks.

Does this mean I should not get the Parallel Computing Toolbox or are there non-GPU accelerations possible there, perhaps with additional cores on the Apple CPU? I see that

https://web.archive.org/web/20241206202254/https://www.mathworks.com/support/requirements/parallel-computing-toolbox.html

"Support for Apple Silicon Macs

Parallel Computing Toolbox now supports Apple silicon Macs with this limitation:

• Distributed and codistributed arrays are not supported for local process pools."

but don't know what that means.

[Damian Pietrus]

The Parallel Computing Toolbox still allows you to start a local parallel pool, meaning that you can run parallel constructs like parfor and parfeval. These are the most common parallel constructs and the ones you'd most likely utilize. For that reason, my personal opinion is that you can still get lots of value from Parallel Computing Toolbox if you're looking to either learn about parallelization or have workflows you're looking to accelerate.

Distributed Arrays are designed for workflows where you have an extremely large array that you need to split up across several parallel workers and still keep in memory. I see this used most often in HPC centers, where a distributed array is split up across workers on several compute nodes. Unless you're working with Big Data or submitting to an HPC cluster, I don't think that limitation is too much of a concern.

[Ravi Narasimhan]

Ok. May I assume tall arrays are supported regardless of the CPU? My objective is to gain experience and experiment so Parallel Computing Toolbox could still be useful. Looks like Python and Julia have some ability to use the Mac GPU so I can try them for that although it will take more tinkering with various Github packages.

I'd like to leave the question open for now in case other Mac-based Home users have any comments on compatibility issues they've encountered.

[Damian Pietrus]

Tall arrays are still supported, which I've just verified on my personal macOS M4 machine.

[Walter Roberson]

Note that it is not clear that Mathworks will ever support Apple's GPU. That is not a "definitely not", but to the best of my knowledge Mathworks is still monitoring the situation with no concrete plans to support Apple GPU.

[Ravi Narasimhan]

Your insights have a fine track record. If TMW is monitoring the situation after many years and iterations of the M-series processors, I will take that as a sign. I can try the alternatives to learn how to interact with a Mac GPU and if it is really necessary, call the Python or Julia code from MATLAB.

Aside: Is it possible to quote prior posts on this Forum? I thought it was in the past but don't see any options for that now.