I was lead programmer for a game company in Dallas, and I wrote nearly all of our GPU bound code... We write a tiled lighting system completely in OpenCL, and our game was written from scratch in c++. I also wrote most of our data structures like lists, maps, and thread safe queues in c.
I am available to work on this now, and depending on how much code it is, think it should take a couple days to complete the optimization pass.