Gpu kernel programming just got a memory-safe path that keeps the speed. nvidia's cuda rust,…
gpu kernel programming just got a memory-safe path that keeps the speed. nvidia's cuda rust, announced september 10, lets you write gpu kernels in rust without losing performance, making the worst c++ bug class a compile error.
the last unsafe frontier in systems programming is closing.
Context
NVIDIA's technical blog post of September 8, 2026 introduces two Rust routes to CUDA, and not one language. cuda-oxide is a custom rustc codegen backend that compiles SIMT-style GPU kernels in Rust to PTX, needs a pinned nightly toolchain and LLVM, and the post says it remains in early alpha. cutile-rs is tile-based GPU programming in stable Rust 1.89 or later with CUDA 13.3, published on crates.io. The NVlabs repository calls cuda-oxide an experimental Rust-to-CUDA compiler and safe(ish), under Apache License 2.0.
The note says announced September 10, and the blog post is dated September 8, 2026. The post says both projects enforce memory safety at compile time, but the examples concern aliasing and out-of-bounds access, and a launch is checked at launch time and not purely at compile time. So the worst C++ bug class becoming a compile error is the note's gloss. No benchmark was found in the passages read, so without losing performance is not verified for these projects. The NVlabs repo corroborates the project and not every claim in the blog, and the blog's attribution was not independently settled.
Watch next
- A cuda-oxide release moving off alpha, and benchmarks against C++ CUDA kernels.
Sources
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 21 September 2026 at 22:46 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →