I of course hope that CUDA will perform as expected and yield useful science. But if it does not, it of course also does not make sense to spend the money on the electricity required to keep the graphics cards crunching, and it would make sense just to focus on the CPU work.
If Dr. Pande would please provide an update as testing progresses regarding how CUDA is performing science-wise, that would be most appreciated!

(And it is interesting that the GPU's are a supplement to, and not a replacement for, the CPU work. So running CPU's continues to be very productive!
