NVIDIA AI Just Released cuda-oxide: An Experimental Rust-to-CUDA Compiler Backend that Compiles SIMT GPU Kernels Directly to PTX

Short summary

NVIDIA released cuda-oxide v0.1.0, an experimental Rust-to-CUDA compiler backend enabling developers to write GPU kernels in Rust. It compiles #[kernel]-annotated functions through stable MIR → Pliron IR → LLVM IR → PTX. Single cargo command handles both host and device code.

• •NVIDIA NVlabs released cuda-oxide v0.1.0, an experimental Rust compiler backend
• •Compiles Rust GPU kernels directly to CUDA PTX bytecode
• •Single cargo command compiles both CPU and GPU code together

Generated with AI, which can make mistakes.