One of the things LabVIEW is best at is its innate parallelism. Parallelizing for loops with a right-click is something other languages wish they had. Having all code on the block diagram (or equivalent) innately run in parallel is something few other languages have even tried, much less succeeded at.
Crunching large datasets can be sped up through use of GPUs, and over the past decade or more, Nvidia has kept their promise of having a common interface to their hardware through CUDA.
Having a LabVIEW add-on that is well maintained, and that can give GPU parallelism in the same ways we've seen LabVIEW deliver CPU parallelism would be a game changer for many labs and manufacturing environments. It could also help LabVIEW be a leader in the AI space.
I suggest that you create this add on package using CUDA as the underlying GPU calls, in order to keep the code easy to manage, while also providing the package for a large number of supported GPUs.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.