What is the difference in terms of performance improvement for Parallel Computing Toolbox in R2013a and R2013b?

Hi All,

Currently my Matlab programs are based on R2011b + Parallel Computing Toolbox 5.2, and I'm planning to upgrade them to GPU based parallel processing. Now I have two versions R2013a and R2013b, with Parallel Computing Toolbox 6.2 and 6.3 respectively. Whichever will be great to have to exploit all sorts of performance improvement?

I searched around mathworks and get some tips from following links

http://www.mathworks.com/help/distcomp/release-notes.html#btw2adn-1 http://www.mathworks.com/support/bugreports/search_results?utf8=%E2%9C%93&search_executed=1&keyword=&release_filter=Fixed+in&release=238&selected_products=6739&commit=Search)

I will be happy to know more about it from all the experts around the world.

Thx Anurag