CUDA's future on the Mac OS X platform

I would like to know what the general sentiment is regarding CUDA’s future on the Mac OS X platform.
Most of my work/research is based upon CUDA running on Mac OS X and I’m starting to wornder if I’m standing on solid ground.

As we all know the latest Mac Pro workstation from Apple doesn’t offer NVIDIA as an option. Now, I know Apple has done that in the past and it has always supported NVIDIA video cards down the road, but still. I also know Apple created OpenCL so I don’t expect them to be overly enthusiastic about CUDA but as long as NVIDIA and its partners do their part we should be fine.

However there are no Tesla video cards for Mac, the most recent CULA Tools don’t offer Mac versions (they are on the way I suppose), there seems to be no Mac version of the newly released cuDNN Toolkit and I’m starting to see a trend that worries me.

Sometimes as a programmer it’s easy to miss the big picture so I could definitively use some advice.



Observation: I have a mid 2009 MacBook Pro (yep it’s old) with two GPU’s (Geforce 9600 and 9400) recently upgraded to YOSEMITE (osX 10.10) I was concerned about graphics driver updates for my “massive” 32 CUDA core capability…NVIDIA CUDA update is now part of System -> Preferences -> CUDA! (in osX)

I am really pleased with this “reach-back” support on what would be considered deprecated technology. I suspect power considerations (by Apple) will drive future acceptance. It’s my understanding NVIDIA is going to great lengths to decrease power consumption and increase “efficiency”.

I was disappointed to see the current GPU options on MAC Pro…I think whomever has the best tools to convert/modify existing code to accommodate Massively Paralleled Processing transition is going to be around for the foreseeable future.