How to reduce the number of workers in the backgroundPool?

I would like to limit the backgroundPool to 2 workers. But when I restart my computer, then open MATLAB, then run the following code I get 10 workers. Is there anyway to reduce the number of backgroundPool workers? Thanks in advance!

Input:

maxNumCompThreads(1)
backgroundPool

Output:

>> maxNumCompThreads(1)
ans =
     6
>> backgroundPool
ans = 
 BackgroundPool with properties: 

           NumWorkers: 10
                 Busy: false

[Rick Amos]

The only way to reduce the number of backgroundPool workers as of R2023b is to start MATLAB with a single computational thread via -singleCompThread startup option.

Could I ask what you had in mind that needs a background pool of 2 workers?

In addition, I see you get 10 backgroundPool workers when maxNumCompThreads at MATLAB startup returns 6. This is a bug, you should be getting 6 backgroundPool workers. Does your machine have an Intel 12th gen chip with P-cores and E-cores? If so, this should at least go back to 6 in a near future version of MATLAB.

[James London]

Good to know! Thanks!

I am using an Intel i5-12600KF processor which I belive has 6 P-cores and 4 E-cores.

I am performing live image processing from a custom microscope. I am doing the image processing on the backgroundPool. This allows a user of the microscope to use a GUI to adjust the movemnt of the microscope in real time while images a being processed. The processing does take up a fair amount of RAM but is very fast. I want to limit the number of backgroundPool workers to prevent all of the RAM from being used by the backgroundPool so that new images can come in and be queued up for processing.

[Walter Roberson]

If you parfeval() exactly two functions, then the memory will not build up.

Answer 1

[Monisha]

delete(gcp('nocreate')); % Delete any existing parallel pool parpool(2); % Create a new parallel pool with 2 workers

This code will delete any existing parallel pool and create a new one with only 2 workers. Place this code at the beginning of your script or function to ensure that it's executed whenever MATLAB starts. This should help you maintain the desired number of workers even after restarting your computer and reopening MATLAB.

[Walter Roberson]

The above will not create a background pool.