I thought that OpenCL was supposed to be the "open" standard for GPU computing... why is everyone rallying around CUDA instead? (seemingly)?
marketing.
Easier to program (at least for me it is easier to understand a piece of CUDA code) and there is a lot of good documentation for CUDA.
Another plus for CUDA is that it lets you use C++.
given that OS X ships with OpenCL, which already runs on nvidia hardware, and is open, and portable, i don' t understand why anyone would want to switch to CUDA.