Actually, in looking at the source recently, I believe I saw a reference in the #include section to OpenCL. Hang on a sec and I'll check that.

EDIT: Yeah. Here it is:

#ifdef OCL

#include <CL/opencl.h>

#endif

At first I thought it was referencing OpenGL! Haha! I almost panicked then because I'm altering it slightly to work on jailbroken iPod Touch/iPhone and the hardware for those devices has absolutely no support for OpenGL (that I know of). Then I realised that it said "opencl.h" and not "opengl.h" and then calmed down a bit.

That's great news! I know you have the skills to be able to do this, I never could :p.


If used to calculate the physics, yes depending on how good your your GPU is (assuming that's what the OpenCL will be written optimized to use) possibly a 50 times speedup, if coded for using a few CPU cores then a few times speedup. The best part is most of the game uses single precision floats or integers which GPU's do VERY well with compared to doubles.

Maybe you could write a super nice email to AMD and ask them how they did it? ^^

Using a gpu instead of cpu can get you around 4000 times more fps if you use a nvidia phyx card, I wont notice a difference though I have no lag what so ever