Question

I'm composing a pretty heavy duty octave function that needs to be minimized with a function such as fminunc (technically I can constrain the variables, but I need to minimize over multiple variables).

A cursory reading of google and Octave documentation suggests that this is a single-threaded process. Are there any options to utilize multi-core/threaded processing here?

I'm on a 12 core box, it feels sad to leave 91% of the compute capacity unused.

Matlab appears to have something useful in this respect with fmincon:

http://www.mathworks.com/help/optim/examples/minimizing-an-expensive-optimization-problem-using-parallel-computing-toolbox.html

Was it helpful?

Solution

This looks like the answer I actually needed. The minimizer may not run multithreaded, but the matrix operations in the function I'm minimizing can.

Get GNU Octave to work with a multicore processor. (Multithreading)

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top