#cuda-kernels Repositories
Discover 34 active open source GitHub repositories tagged with #cuda-kernels, accumulating 129,157 total stars (averaging 3,798 stars per project). Leading this topic is Dao-AILab/flash-attention with 25,330 stars.
34 repositories total
An Open-source book with Modern CUDA learning notes for beginners โ BF16/FP8/FP4, HGEMM, CuTe, Flash-Attention, etc.
Hackable and optimized Transformers building blocks, supporting a composable construction.
Ecosystem of libraries and tools for writing and executing fast GPU code fully in Rust.
[MLSys 2024 Best Paper Award] AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration
A project demonstrating Lidar related AI solutions, including three GPU accelerated Lidar/camera DL networks (PointPillars, CenterPoint, BEVFusion) and the related libs (cuPCL, 3D SparseConvolution, YUV2RGB, cuOSD).
[ECCV 2024] Street Gaussians: Modeling Dynamic Urban Scenes with Gaussian Splatting
Official Pytorch repository for Extreme Compression of Large Language Models via Additive Quantization https://arxiv.org/pdf/2401.06118.pdf and PV-Tuning: Beyond Straight-Through Estimation for Extreme LLM Compression https://arxiv.org/abs/2405.14852
A machine learning library with a TypeScript API and Rust backend. CUDA and WebGPU compatibility. Built to understand how ML frameworks and models work internally.
cuDNN Frontend is NVIDIA's modern, open-source entry point to the cuDNN library and a growing collection of high-performance open-source kernels.
A Distributed Attention Towards Linear Scalability for Ultra-Long Context, Heterogeneous Data Training
An Axum-based Rust web framework with Lean 4-verifiable application kernels that run on servers, bare-metal embedded devices, and GPUs, plus integrated security testing.
[MLSys'25] QServe: W4A8KV4 Quantization and System Co-design for Efficient LLM Serving; [MLSys'25] LServe: Efficient Long-sequence LLM Serving with Unified Sparse Attention
FlashRT is a high-performance realtime inference engine for small-batch, latency-sensitive AI workloads. The flagship integration is production VLA control for Pi0, Pi0.5, GROOT N1.6, and Pi0-FAST. Also support llm e.g, qwen3.6-27B
Pure Rust + CUDA LLM inference engine โ no PyTorch, OpenAI-compatible, serves Qwen3 to Kimi-K2
Open source impl of **MV-DUSt3R+ Single-Stage Scene Reconstruction from Sparse Views In 2 Seconds** from Meta Reality Labs. Project page https://mv-dust3rp.github.io/
CUDA-L2: Surpassing cuBLAS Performance for Matrix Multiplication through Reinforcement Learning
[ICCV 2025] A simple training-free approach adapting DUSt3R for dynamic scenes.
Frequently Asked Questions
What is the most popular #cuda-kernels repository on GitHub?
The top ranked open source #cuda-kernels project on GitHub is Dao-AILab/flash-attention with 25,330 stars. Fast and memory-efficient exact attention
Which programming languages are most used in #cuda-kernels repositories?
Among the top ranked #cuda-kernels projects on GitHub, the most common programming languages are Python, C++, Rust, Cuda.
How is the #cuda-kernels repository ranking calculated?
We track 34 active open source repositories tagged with #cuda-kernels (500+ stars), totaling 129,157 stars with an average of 3,798 stars per repository. Rankings update daily from the GitHub API.