Does matlab 2012a auto vactorize?

1 view (last 30 days)
I recently had 2012a installed at work. I had previously had 2010. I noticed a surprising performance improvement in code that contained for loops, that had not been vectorized. Several people at work noticed this causing us to speculate that matlab contained an internal vectorized version of "for".
Can anyone shed light on this? Thanks.

Accepted Answer

Walter Roberson
Walter Roberson on 28 Jul 2012
Yes, between those two releases, MATLAB did make a significant optimization improvements. One of the significant improvements was that MATLAB now automatically predicts preallocation requirements, saving significant time in memory management if the programmer had not put in preallocation.

More Answers (1)

David Bergman
David Bergman on 28 Jul 2012
Walter, thanks. Do you know of any documentation on this? It makes me wonder if it's even worth learning to vectorize ;-). Once you get good at it, it makes you thinks differently about how to structure and write code. Many young engineers are educated on matlab and never learn C/C++ or FORTRAN. If one writes C-style code in matlab can you expect it to run just as fast as C? Thanks again.
  1 Comment
Star Strider
Star Strider on 28 Jul 2012
For documentation, I refer you to ‘Release Notes’: http://www.mathworks.com/help/techdoc/rn/rn_intro.html.
For the rest I refer you to Walter Roberson.

Sign in to comment.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!