Particle Filters

[Old posts from the commercial version of ArrayFire] Discussion of ArrayFire using CUDA or OpenCL.

Moderator: pavanky

Particle Filters

Postby melonakos » Tue Jul 26, 2011 6:14 pm

We recently received the following question by email:

My lab is thinking of venturing into GPU computing. We use particle filters and the dimensionality of the problems we will be studying will be increasing, so our hope is that significant savings will be found through parallelization.

Our interest is in libJacket. Have you run any benchmark tests on such problems? There is minimal discussion of particle filtering coding on your forum, so it is difficult to get a sense of whether the library is appropriate for the task. Thanks.


I don't think we have any canned examples/benchmarks of particle filters. But we have had a lot of computer vision customers mention success in similar methods.

Particle filters rely heavily on matrix manipulations, arithmetic in the functions, normailization, recalculation of distributions, reductions, etc. Depending on the problem, you may consider putting just the functional on the GPU or putting the entire filtering process there as well.

Building block function benchmarks are provided here: http://accelereyes.com/products/benchmarks_libjacket

Good luck!
John Melonakos (john@arrayfire.com)
User avatar
melonakos
 
Posts: 503
Joined: Tue Jun 10, 2008 9:49 am

Return to [archive-commercial] Programming & Development with ArrayFire

cron