MATLAB GPU toolbox: PCT, GPUMat and Jacket revisted

· Programming
Authors

In my last benchmark of the MATLAB GPU toolboxes, I compared PCT from The MathWorks, Jacket from Accelereyes and GPUMat from GPYou. I now updated theses results with new hardware and new some more optimized code.

Previously, we used a GeForce GTX 275. Now, we also use the GeForce GTX 520 Ti.

Click to enlarge

On this hardware, the GPUMat dramatically improves: For large datasets, GPUMat on a GeForce GTX 520 Ti now almost as good as Jacket non-compiled. On the GTX 520 Ti, Matlab PCT improves (2x to 3x) while Accelereyes Jacket worsens (14x to 12x). But, Jacket is still the fastest GPU toolbox. For the complete report see the updated previous post on benchmarking MATLAB GPU toolboxes.

1 Comment

Comments RSS
  1. Ryan G

    I’m surprised a MATLAB monte carlo simulation did better on Jacket. I believe the PCT toolbox now supports gpu which may increase performance of that tool.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s