Cuda Repositories
Explore 65 active open source Cuda repositories on GitHub with a combined 165,484 stars (averaging 2,545 stars per project). The #1 ranked Cuda project is karpathy/llm.c with 30,323 stars.
65 repositories total
The project is an official implementation of our CVPR2019 paper "Deep High-Resolution Representation Learning for Human Pose Estimation"
[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.
[MICRO'23, MLSys'22] TorchSparse: Efficient Training and Inference Framework for Sparse Convolution on GPUs.
[ICML2025] SpargeAttention: A training-free sparse attention that accelerates any model inference.
Distributed training & inference for multimodal perception: shared embeddings across text, images, video, audio, code & graphs from 64 to 1024 dimensions ๐ผ๏ธ
RAFT contains fundamental widely-used algorithms and primitives for machine learning and information retrieval. The algorithms are CUDA-accelerated and form building blocks for more easily writing high performance applications.
cuDNN Frontend is NVIDIA's modern, open-source entry point to the cuDNN library and a growing collection of high-performance open-source kernels.
Training materials associated with NVIDIA's CUDA Training Series (www.olcf.ornl.gov/cuda-training-series/)
Examples demonstrating available options to program multiple GPUs in a single node or a cluster
HPC tutorial, covering collective communication (MPI, NCCL), CUDA programming, vectorization SIMD, RDMA communication, etc.
Additional utils and helpers to extend TensorFlow when build recommendation systems, contributed and maintained by SIG Recommenders.
Python library for Room Impulse Response (RIR) simulation with GPU acceleration
NVIDIA NVSHMEM is a parallel programming interface for NVIDIA GPUs based on OpenSHMEM. NVSHMEM can significantly reduce multi-process communication and coordination overheads by allowing programmers to perform one-sided communication from within CUDA kernels and on CUDA streams.
Frequently Asked Questions
What is the most starred Cuda repository on GitHub?
The most starred active Cuda repository on GitHub is karpathy/llm.c with 30,323 stars. LLM training in simple, raw C/CUDA
How many popular Cuda projects are tracked on Git Stars?
We currently track 65 active open source Cuda repositories with 500+ stars, accumulating 165,484 total stars and an average of 2,545 stars per project.
How often is the Cuda GitHub ranking updated?
All Cuda repository star counts, activity dates, and rankings are synchronized daily from the official GitHub API.