Certainly. Have a look here: https://github.com/peterkutz/GPUPathTracer. It's a physically based path tracer, I it uses a kd-tree but I am not sure if they actually got around to implementing it. Well, actually, CUDA already has kd-tree code (cukd). There's also tokaspt, which is a CUDA port of the "smallpt" program. An OpenCL port also exists, but has less features.
As for whitted ray tracing, though, I don't know of any, but that's probably because I don't look around for that. You should be able to modify the code to a whitted model fairly easily, though, the hardest part in a ray tracer is getting the initialization logic and various subsystems (ray-scene intersection, asset loading) up, the actual rendering code is very simple.
Plus, why would you implement whitted ray tracing on a GPU when rasterization can achieve pretty much the same visual results much faster? I believe that is why people prefer to implement realistic rendering methods on the GPU, rather than basic ray tracing, but there have been whitted benchmarks published so there must be some code somewhere.
Also, welcome to GameDev!
Edited by Bacterius, 13 February 2013 - 01:34 AM.
The slowsort algorithm is a perfect illustration of the multiply and surrender paradigm, which is perhaps the single most important paradigm in the development of reluctant algorithms. The basic multiply and surrender strategy consists in replacing the problem at hand by two or more subproblems, each slightly simpler than the original, and continue multiplying subproblems and subsubproblems recursively in this fashion as long as possible. At some point the subproblems will all become so simple that their solution can no longer be postponed, and we will have to surrender. Experience shows that, in most cases, by the time this point is reached the total work will be substantially higher than what could have been wasted by a more direct approach.
- Pessimal Algorithms and Simplexity Analysis