Does maltab parameter estimation app use CPU cores or GPU cores ?

Does maltab parameter estimation app uses CPU cores or GPU cores ?

I'm asking this question because I'm running parameter estimation with huge amount of data for accuracy, but I'm running out of memory during serialization .

And a friend told me he has a 3 gtx1070 cuda instances. He is asking me either matab parallel option available in parameter estion uses CPU or GPU cores. Please can anyone give me details about this ?

[KODJOVI SENAM BORIS AHIANDIPE]

Anyone to explain me this please ?

Answer 1

[Sumukh]

Hi Kodjovi,

The parameter estimator app uses either CPU or GPU cores depending on the availability of GPUs and configuration of the system used for parallel computing. To use the parallel computing capability, a parallel pool must be started before using the estimator app. When the command “parpool” is called to start the parallel pool of workers with the default preferences, one worker is assigned to one physical core of the CPU. You can refer to the following documentation to understand more about this:

https://www.mathworks.com/help/releases/R2023a/parallel-computing/parpool.html

The command assigns the workers to the GPUs instead of the cores on the CPU if they are available for use in MATLAB. You can refer to the following documentation to determine if the GPUs available for use in MATLAB and use them in parallel pool:

https://www.mathworks.com/help/releases/R2023a/parallel-computing/parpool.html#mw_067600d3-2012-4b07-913d-510d96d492ec

Additionally, you can refer to the following documentation to troubleshoot the configuration steps for using parallel computing in parameter estimation app:

https://www.mathworks.com/help/releases/R2023a/sldo/ug/how-to-use-parallel-computing-pe-gui.html#br1wi9a-1

I hope this answers your query and helps resolve the issue.