Hello fellow CUDAers,
Once again, a release of my cuda-api-wrappers library is coming up: Version 0.5.2; there’s a list of changes at the link. In particular, I’ve made multiple changes in recent versions to streamline the facilities for dynamic compilation of programs (NVRTC-like functionality).
If you have experience with NVRTC, I would ask for your help in trying out the library to compile some programs, and giving feedback, telling me what may be missing or inconvenient to use.
I’ve also asked for some feedback recently on a launch configuration builder class - and got some. You can now try out the result, e.g. as in this simple example program.
More generally, the 0.5.x line of releases finally fuses the functionality of the runtime and driver API - hopefully in a seamless fashion. I’m also interested in feedback on whether I’ve pulled this off convincingly enough.
Thanks goes to the many people who have starred so far, and even greater thanks to people who have filed bugs - those are always educational and useful.