Integrate OpenCL data-parallel acceleration as core part of Bullet

[GoogleCodeExporter]

Several experimental projects have been targeting GPU Physics in the past:

1) Steve Baker's shader based GPU collision/physics
2) CUDA broadphase and constraint solver
3) OpenCL broadphase and simplified pipeline

This data parallel work should be properly integrated into Bullet 3.x using 
OpenCL.

You can follow progress in branches/OpenCL

Original issue reported on code.google.com by erwin.coumans on 21 Jan 2010 at 7:58

[GoogleCodeExporter]

Steve Bakers work started in 2006, see the thread here:
http://bulletphysics.org/Bullet/phpBB3/viewtopic.php?f=9&t=500&hilit=steve+baker
And the implementation in Bullet/Extras/obsolete/GPUphysics

CUDA work together with Takahiro Harada was shown at GDC 2009, see attached 
slides
See http://bulletphysics.org/wordpress/?p=64

The OpenCL efforts was presented at NVidia GPU conference 2009,
http://www.nvidia.com/content/GTC/documents/1077_GTC09.pdf
or see attached slides

Or browse the OpenCL branch here:
http://code.google.com/p/bullet/source/browse/#svn/branches/OpenCL

Original comment by erwin.coumans on 21 Jan 2010 at 8:03

Attachments:

• 1077_GTC09.pdf

[GoogleCodeExporter]

Focus is shifting to Bullet 3.x so new features are moved to a new issue 
tracker:

Split into several issues:

OpenCL parallel constraint solver
https://github.com/erwincoumans/experiments/issues/5
OpenCL parallel narrow phase convex collision detection
https://github.com/erwincoumans/experiments/issues/6
OpenCL parallel broadphase collision detection
https://github.com/erwincoumans/experiments/issues/7
OpenCL parallel midphase (triangle mesh) collision detection
https://github.com/erwincoumans/experiments/issues/8

(most updates to Bullet 2.x will be bug fixes and maintenance)

Original comment by erwin.coumans on 1 Nov 2011 at 4:33

• Changed state: WontFix