Will Matlab mldivide( \ ) operator use all the available cores when used on multi-node?

I am trying to solve a transient problem in matlab in which I have to solve x = A\b for every iteration. A is a sparse array (not distribured) of size around 2millionx2million and 350MB. Matlab documentation says that \ operator is optimized and can use all the cores.

I'm currently running on a single node using all 28 cores available to solve the problem and planning on using distributed computing on multiple nodes. I do not know if matlab is using all the available cores on multiple nodes during computation of ( \ ).

Is there anyway to check if the code is using all the cores and if not what can be done to use all the available cores.

Answer 1

[Walter Roberson]

It will use up to the current maxNumCompThreads https://www.mathworks.com/help/matlab/ref/maxnumcompthreads.html

The number of threads per worker can be configured in the cluster profile, or at job submission time https://www.mathworks.com/matlabcentral/answers/395333-using-parfor-on-cluster-changing-number-of-cores-per-worker#answer_315541

[Raj Gopal Nannapaneni]

Thank you for the information. And for the code to use all the computing threads should I change my array to a distributed array or does matlab takes care at the backend?

[Walter Roberson]

MATLAB will figure out the best way to send the array.

[Raj Gopal Nannapaneni]

Thank you