This package contains 5 functions.
The lowest-level function in the package, parProcess, interacts with the undocumented remoteparfor object in a manner similar to the PCT PARALLEL_FUNCTION, but only for the supply-process-consume mode of operation. parProcess is less constrained than PARALLEL_FUNCTION in that the supplier can decided dynamically what data to send for processing, and when to stop processing.
-- Two applications of parProcess are provided.
pforfun provides a PARFOR like functionality in a function call. Unlike PARFOR, the scheduling is somewhat configurable, and dynamic load balancing is provided. A progress bar is also available.
parallel_function can be used to mask the PCT PARALLEL_FUNCTION used by PARFOR. This allows the user to redirect parfor usage to parProcess. Similar to pforfun, scheduling is configurable and a progress bar is available as an option.
This is a very simple example of how pforfun works. It can be run with no arguments, to test pforfun and parProcess. Run this after running MATLABPOOL.
A slightly more complicated version of matlabpool('size').
NOTE: All of this rests on undocumented MATLAB code, so you should expect it to work only on R2009a! Of course, one might hope that things won't change too much in the near future...
Download apps, toolboxes, and other File Exchange content using Add-On Explorer in MATLAB.