Matlab functions are executed locally and not on the cluster

Hello, I have a pc A which is my client node, a computer B which is a job master and 4 computers (C, D, E, F) that host 16 workers. Using admincenter, I've configured a parallel cluster which validates correctly on pc A. On A, I launch a computation by writing

matlabpool
myscript

now, only explicit parallel constructs (like parfor) are executed on the cluster. Everything else runs on A, including an A \ b statement wich uses all of the 4 cores available on A (which is a macbook pro). Nothing should be executed on A and everything (even non-parallel statements) should be run executed on the cluster

Computers B,C,D,E and F are linux workstations (ubuntu 12.04). What am I missing?

[Daniel Shub]

I am not 100% sure, but just because something like A\b can use multiple cores/threads, doesn't mean it can use multiple machines/processes.

[Patrik Osgnach]

Makes no sense, Matlab is distributed in its very own nature. But this is beyond the point. I need those statements to be executed on the cluster, not on the client pc

Answer 1

[Jason Ross]

If you want to have those statements executed on the cluster, you need to use a parallel constructs like batch, smpd, parfor or the job/task interface.

There is an exmaple of A/b on the cluster that uses spmd: http://www.mathworks.com/help/distcomp/examples/benchmarking-a-b.html