nvrtc

NumPy & SciPy for GPU

Safe rust wrapper around CUDA toolkit

A single-header C++ library for simplifying the use of CUDA Runtime Compilation (NVRTC).

Hooked CUDA-related dynamic libraries by using automated code generation tools.

Pythonic particle-based (super-droplet) cloud microphysics (and aqueous-chemistry) package with box, parcel & 1D/2D prescribed-flow examples in Python, Julia and Matlab

CUDA tool set for non-C++ languages that provides similar functionality like Thrust, with NVRTC at its core.

Fractal Explorer with an ability to write custom formulas

Yet Another CudaExecutor - wrapper to easily compile and execute cuda kernels

A from-scratch nano-Triton and nano-Helion: the modern GPU-kernel DSL stack, rebuilt in ~4,000 lines

CURandRTC is a GPU random number generation module based on ThrustRTC.

NVRTC bindings for RUST

A CUDA interface for Python. A distillation of the engine part of ThrustRTC.

Example of using NVIDIA's runtime CUDA compilation API and JIT linking

CUDA Runtime Compilation and Dynamic Parallelism samples.

GPU-first spatial analytics for Python. Drop-in GeoPandas replacement powered by runtime-compiled CUDA kernels

Math DSL compiler with evaluation on GPU.

Actor-shaped face for compute acceleration backends (NVIDIA CUDA shipping; ROCm / Metal / oneAPI / Vulkan future). Built on the atomr actor runtime.

Experimental no-addon Node.js CUDA runtime and toolchain; Linux-reference architecture with exact Windows public-alpha evidence.

Native Rust engine for neural network inference and training: LLM, image, video, speech and music models on hand-written CUDA kernels JIT-compiled via NVRTC for any NVIDIA sm_80+ GPU (native NVFP4/MXFP8 on Blackwell sm_120+). .syn bundles in NVFP4/MXFP8/SQ1–SQ8, GGUF loads directly. No PyTorch, no Python.

C arena and graph engine with owned memory, plus a matched CUDA (NVRTC) population lifecycle: 112 paired CPU/GPU trials identical, measured CPU/GPU crossover and scaling.