Hey Folks!
Now that LTC is gaining some momentum, I'm curious if there's any OpenCL developers out there want to take a stab at optimizing the OpenCL core for scrypt that is used today by cgminer and the ilk. I know with some debug work the bitcoin folks were able to squeeze performance optimizations of 5% or greater by dropping clock cycles required to perform operations.
Anyone out there with more knowledge on the topic than I up for the challenge?