First of all, I'd say this is totally -on- topic =). I haven't worked with CUDA, but I've contemplated messing with it, and I've read enough articles on it to be familiar with some of its advantages and problems.
The biggest comment I would have about it is that it often has odd platform-dependent bugs (even though its primarily for NVIDIA hardware, which is supposed to be pretty consistent as far as hardware goes). Things like if-else often don't work right on all GPU's, and you wind up having to rephrase your statements in awkward or verbose ways to get things working. Other than that, it's certainly better than trying to figure out how to translate a problem into OpenGL, for instance. Depending on the problem domain, the speedups can be quite dramatic. On Mon, 2009-03-16 at 22:03 -0400, J. Milgram wrote: > Anybody here doing anything with CUDA? > > Judah
