undefined | Better HN

0 pointsmike_hearn1y ago0 comments

You don't need to use C++ to interface with CUDA or even write it.

A while ago NVIDIA and the GraalVM team demoed grCUDA which makes it easy to share memory with CUDA kernels and invoke them from any managed language that runs on GraalVM (which includes JIT compiled Python). Because it's integrated with the compiler the invocation overhead is low:

https://developer.nvidia.com/blog/grcuda-a-polyglot-language...

And TornadoVM lets you write kernels in JVM langs that are compiled through to CUDA:

https://www.tornadovm.org

There are similar technologies for other languages/runtimes too. So I don't think that will cause NVIDIA to lose ground.

0 comments

gymbeaux1y ago

So these alternatives exist yes, but are they “production ready”- in other words, are they being used. My opinion is that while you can use another language, most companies for one reason or another are still using C++. I just don’t really know what the reason(s) are.

I think about other areas in tech where you can use whatever language, but it isn’t practical to do so. I can write a backend API server in Swift… or perhaps more relevant- I can use AMD’s ROCm to do… anything.

j / k navigate · click thread line to collapse