I was excited to see that there was a Mac OS X “CUDA Toolkit” specifying Xcode 7.3. I installed this older version of Xcode and downloaded this Toolkit, and now I’m scratching my head wondering where the “comprehensive development environment for C and C++ developers” is, and wondering if this technology is appropriate for my use case. I don’t see a single Xcode project file in this toolkit? There really isn’t anything in the documentation PDFs that shows or hints at how to use CUDA from Xcode either. There is a “samples” folder of command-line demos built with makefiles, but I’m not making demos or doing scientific computing, I’m working with on a large, existing image-processing application that I would like to add GPU acceleration to. It’s an in-house tool, but a full GUI application built from an Xcode project using the Xcode IDE, mostly Objective-C source, and used by artists on varying configurations of Macs (but all with Nvidia hardware, mostly GT 750m Macbooks).
So my real question is, for this kind of application development, is this CUDA toolkit even something I can make use of? It’s not the Xcode-7.3-ready SDK I was hoping for, but is there some relatively easy way to build some CUDA-accelerated code into a dynamic library and then link it into my app? Or am I completely barking up the wrong tree here, should I be looking at implementation through OpenCL instead on the Mac OS platform?