I am running compiled Matlab code on a shared cluster. As such, it is important that I don't overuse my share of the computational resources. However, I am failing at being able to control the amount of computational resources that Matlab uses when run on a 48 core shared memory machine - on one recent run, a single process used over 1245% CPU! (So, about 13 cores worth of computing).
I have tried to use the setNumMaxThreads option, which does not work. I have tried to compile with -singleCompThread option (that is opening Matlab when I originally compile using matlab -singleCompThread). I do not have access to the parallel computing toolbox (my school will not let small clusters purchase it.)
Does anyone have an idea of how I can stop this??