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

DeepGEMM in pure Rust: FP8/BF16 fine-grained-scaled GEMM + MoE grouped GEMM for NVIDIA SM90a (TMA+WGMMA warp-specialized kernels, NVRTC runtime JIT, zero torch/nvcc dependency)

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.

CUDA Runtime Compilation and Dynamic Parallelism samples.

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

Header-only CUDA C++17 library for rigorous interval and affine arithmetic, with guaranteed error bounds (nvcc and NVRTC)

Eigen 5.0.1 usable in CUDA kernels compiled at runtime with NVRTC

Math DSL compiler with evaluation on GPU.