CUDA (Compute Unified Device Architecture) is a parallel computing platform and application programming interface (API) model created by NVIDIA for general computing on its own GPUs (Graphics Processing Units). It empowers application developers to leverage the parallel processing capabilities of NVIDIA's GPUs to accelerate computation-heavy tasks, such as matrix operations, physics simulations, deep learning training, and real-time video processing. CUDA provides a C-like programming language that allows developers to write kernel functions, which are executed on the GPU, and manage memory between the host (CPU) and device (GPU) environments. Utilizing CUDA can lead to significant performance improvements in suitable applications, and it integrates well with various programming environments, including Python through libraries like PyCUDA or through frameworks like TensorFlow with GPU support. Understanding basic concepts such as kernels, threads, blocks, and warps is essential for developers to effectively harness the power of GPU programming with CUDA.
View the most prominent open source CUDA projects in the list below. Click on a specific project to view its alternative or complementary packages. Make comparisons and find the best package for your app.
- A high-throughput and memory-efficient inference and serving engine for LLMs☆69,622Updated this week
- World's fastest and most advanced password recovery utility☆25,294Updated 2 months ago
- Build and run Docker containers leveraging NVIDIA GPUs☆17,484Updated 2 years ago
- SGLang is a high-performance serving framework for large language models and multimodal models.☆23,091Updated this week
- Instant neural graphics primitives: lightning fast NeRF and more☆17,236Updated last month
- kaldi-asr/kaldi is the official location of the Kaldi project.☆15,317Updated 4 months ago
- CUDA on non-NVIDIA GPUs☆13,888Updated last week
- Open3D: A Modern Library for 3D Data Processing☆13,301Updated last week
- Burn is a next generation tensor library and Deep Learning Framework that doesn't compromise on flexibility, efficiency and portability.☆14,227Updated this week
- Solve puzzles. Learn CUDA.☆11,932Updated last year
- TensorRT LLM provides users with an easy-to-use Python API to define Large Language Models (LLMs) and supports state-of-the-art optimizat…☆12,811Updated this week
- NumPy aware dynamic Python compiler using LLVM☆10,884Updated this week
- NumPy & SciPy for GPU☆10,762Updated this week
- OneFlow is a deep learning framework designed to be user-friendly, scalable and efficient.☆9,390Updated 2 months ago
- cuDF - GPU DataFrame Library☆9,473Updated last week
- Containers for machine learning☆9,221Updated last week
- A fast, scalable, high performance Gradient Boosting on Decision Trees library, used for ranking, classification, regression and other ma…☆8,781Updated last week
- CUDA Templates and Python DSLs for High-Performance Linear Algebra☆9,226Updated this week
- Samples for CUDA Developers which demonstrates features in CUDA Toolkit☆8,801Updated last month
- Modular ZK(Zero Knowledge) backend accelerated by GPU☆7,728Updated last year
- Go package for computer vision using OpenCV 4 and beyond. Includes support for DNN, CUDA, OpenCV Contrib, and OpenVINO.☆7,371Updated last month
- 📚LeetCUDA: Modern CUDA Learn Notes with PyTorch for Beginners🐑, 200+ CUDA Kernels, Tensor Cores, HGEMM, FA-2 MMA.🎉☆9,536Updated 2 weeks ago
- A flexible framework of neural networks for deep learning☆5,921Updated 2 years ago
- An interactive NVIDIA-GPU process viewer and beyond, the one-stop solution for GPU process management.☆6,540Updated this week
- A Python framework for accelerated simulation, data generation and spatial computing.☆6,191Updated this week
- ALIEN is a CUDA-powered artificial life simulation program.☆5,347Updated this week
- [ARCHIVED] The C++ parallel algorithms library. See https://github.com/NVIDIA/cccl☆4,998Updated last year
- cuML - RAPIDS Machine Learning Library☆5,113Updated this week
- A PyTorch Library for Accelerating 3D Deep Learning Research☆5,032Updated 2 weeks ago
- Supercharge Your LLM with the Fastest KV Cache Layer☆6,839Updated this week
- ArrayFire: a general purpose GPU library.☆4,854Updated 5 months ago
- Tengine is a lite, high performance, modular inference engine for embedded device☆4,504Updated 11 months ago
- Making it easier to work with shaders☆4,987Updated this week
- Lightning fast C++/CUDA neural network framework☆4,405Updated last month
- HIP: C++ Heterogeneous-Compute Interface for Portability☆4,339Updated this week
- Fast inference engine for Transformer models☆4,274Updated last week
- FlashInfer: Kernel Library for LLM Serving☆4,853Updated this week
- Performance-optimized AI inference on your GPUs. Unlock superior throughput by selecting and tuning engines like vLLM or SGLang.☆4,489Updated this week
- A retargetable MLIR-based machine learning compiler and runtime toolkit.☆3,591Updated this week