Jump to content

  • Log In with Google      Sign In   
  • Create Account

#ActualVilem Otte

Posted 26 March 2013 - 05:51 PM

Although I think you'll find both too slow (at least for games). Custom (specific) solution that is not too general can be a lot faster. I'd point you to http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ where there is published paper + code of high efficient gpu ray tracing (NVidia only).

 

I've adopted the solution in my own library also on AMD and the speed measures are quite close (the thing is you still have to use CUDA on NVidia and OpenCL on AMD, as OpenCL support on NVidia gpus is utter garbage).


#1Vilem Otte

Posted 26 March 2013 - 11:03 AM

Although I think you'll find both too slow (at least for games). Custom (specific) solution that is not too general can be a lot faster. I'd point you to http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ where there is published paper + code of high efficient gpu ray tracing (NVidia only).

 

Although I've adopted the solution in my own library also on AMD and the speed measures are quite close (the thing is you still have to use CUDA on NVidia and OpenCL on AMD, as OpenCL support on NVidia gpus is utter garbage).


PARTNERS