Hi Aubai,
As I can understand, you are trying to limit MATLAB’s memory usage according to the available memory. Using MATLAB workspace in preference option is a possible solution but you have already tried that, and it did not work for you.
Here’s a workaround, you can also try changing the java heap size.
To adjust the heap size:
- On the Home tab, in the Environment section, click Preferences. Select MATLAB > General > Java Heap Memory.
- Select a value using the slider or spin box.
Note: Increasing the heap size decreases the amount of memory available for storing data in arrays.
3. Click OK.
4. Restart MATLAB.
But obviously, I can understand that changing Java heap size might not be a good solution in your use case as it is not possible to adjust the heap size dynamically (at least effectively, there are ways to change it with code, but they require a restart).
Another possible workaround can be to use the “memory” function with a break loop. Here’s an example:
% for this example, I am setting memory threshold as 80
thresholdMemory = 80;
while true
% You can call your entry point function here!
% This part checks memory usage.
[~, memObj] = memory;
availableMemory = memObj.PhysicalMemory.Available;
totalMemory = memObj.PhysicalMemory.Total;
memoryInUse = 100*(totalMemory-availableMemory)/totalMemory;
% break loop if memory usage exceeds the threshold.
if memoryInUse > thresholdMemory
break;
end
end
Learn more about the “memory” function using the following link:
I was also able to find a similar MATLAB Answers question, you can also try using methods suggested here,
You can also try using Operating system specific memory management methods like,
Using Windows job object, this is a bit advanced and requires programming with the Windows API. You can also leverage third party applications for the same.
(this method is not supported by MATLAB officially, so it might be more tedious to work with)
In case of linux, you can use the ulimit command in the shell to limit the resources available. “cgroups” feature can also be used. (you can checkout open-source job schedulers like “slurm workload manager”).
(this method is not supported by MATLAB officially, so it might be more tedious to work with)
I hope this helps, thanks!
