I need assistance setting up a build chain that allows compiling CUDA code as part of a C++ node-gyp project. I'm an experienced node and C++ developer, but am new to CUDA.
I made a minimal C++/node-gyp starter project that has a single function that adds one to each element in a uint8_t* array ([url removed, login to view]). The goal is to replace the C++ implementation with a CUDA implementation.
This task thus entails: (1) Modifying the build chain to compile CUDA code. Must work with MSVC and gcc. I know there are multiple ways to set this up, and I don't know which is best for this scenario.
(2) Writing the example `x += 1` CUDA function and its invocation.
Potentially useful references: * [url removed, login to view] (last commit Sep 2014; badly out of date, had problems compiling)
[url removed, login to view] (a suggested [url removed, login to view] file, which is not quite correct) If you have any question kindly let me know.