News: 0001632478

  ARM Give a man a fire and he's warm for a day, but set fire to him and he's warm for the rest of his life (Terry Pratchett, Jingo)

NVIDIA Releases CUDA-Oxide 0.1 For Experimental Rust-To-CUDA Compiler

([NVIDIA] 103 Minutes Ago CUDA-Oxide 0.1)


A new NVIDIA Labs project is greatly improving the capabilities of using the Rust programming language for developing CUDA kernels for NVIDIA GPUs.

CUDA-Oxide 0.1 released on Thursday as the inaugural release of this experimental Rust-to-CUDA compiler for writing SIMT GPU kernels in a "safe(ish)" manner using the Rust programming language. CUDA-Oxide takes standard Rust code and emits directly to the NVIDIA PTX format.

CUDA-Oxide is designed by NVIDIA engineers for single-source compilation, device-side abstractions, and a Rust-native compilation pipeline using a custom rusc compiler back-end.

While currently experimental, this new open-source project is a big step-up for those wanting to deal with CUDA from native Rust code.

"cuda-oxide is an experimental compiler that demonstrates how CUDA SIMT kernels can be written natively in pure Rust -- no DSLs, no foreign language bindings -- and made available to the broader Rust community. The project is in an early stage (alpha) and under active development: you should expect bugs, incomplete features, and API breakage as we work to improve it. That said, we hope you'll try it in your own work and help shape its direction by sharing feedback on your experience."

Those interested in learning more about CUDA-Oxide or trying out the inaugural v0.1 release can do so via the NVLabs project on [1]GitHub .



[1] https://github.com/NVlabs/cuda-oxide



The little town that time forgot,
Where all the women are strong,
The men are good-looking,
And the children above-average.
-- Prairie Home Companion