Gpu

OpenLake is a distributed object storage engine optimized for GPU-accelerated AI workloads, built in Rust using io_uring for ultra-low-latency,...

whichllm is a CLI tool that automatically detects your hardware (GPU, CPU, RAM) and recommends the best local LLM models from HuggingFace that will...

llm-d is a distributed inference serving stack optimized for high-performance LLM deployment on Kubernetes across diverse hardware accelerators. It...

Selkies is an open-source WebRTC-based remote desktop streaming platform optimized for low-latency GPU-accelerated Linux environments, designed for...

Lemonade is a local AI server that runs optimized LLMs on user GPUs and NPUs (particularly AMD Ryzen AI and Radeon), providing OpenAI-compatible APIs...

meshoptimizer is a C++ library providing algorithms to optimize triangle meshes for GPU rendering efficiency and reduce mesh complexity and storage...

cuda-oxide is an experimental Rust-to-CUDA compiler that enables writing GPU kernels directly in safe, idiomatic Rust without DSLs or foreign...

FlashInfer is a specialized GPU kernel library designed for high-performance LLM inference, providing optimized implementations for attention, GEMM,...

Koharu is a local-first ML-powered desktop application for translating manga, combining object detection, OCR, inpainting, and LLM inference entirely...

TypeGPU is a TypeScript-based toolkit for WebGPU that enables developers to write GPU shaders and compute kernels in TypeScript with advanced type...

TT-NN is a Python and C++ neural network operator library built on TT-Metalium, Tenstorrent's low-level kernel programming model for their custom AI...

Ratty is a GPU-rendered terminal emulator written in Rust that enables inline 3D graphics rendering alongside traditional terminal output, inspired...

NCCL is a specialized GPU communication library that implements collective operations (all-reduce, all-gather, broadcast, etc.) optimized for...

G-Helper is a lightweight system utility that replaces Asus's Armoury Crate software for controlling power profiles, fan curves, RGB lighting, and...

CuPy is a GPU-accelerated array library that provides NumPy and SciPy compatibility for NVIDIA CUDA and AMD ROCm platforms, enabling existing...

BarraCUDA is a multi-architecture compiler that translates CUDA, HIP, and Triton source code into binaries for AMD RDNA GPUs, NVIDIA PTX, Tenstorrent...

LACT is a Linux desktop application for monitoring and controlling AMD, Nvidia, and Intel GPUs, offering detailed system information, overclocking,...

TGFX is a lightweight 2D graphics library optimized for modern GPUs, offering high-performance rendering of text, images, and vector graphics across...

macmon is a real-time system performance monitoring CLI for Apple Silicon Macs (M1–M5) that displays CPU, GPU, ANE power usage, temperature, and...

Transformer Engine is a library for accelerating Transformer model training and inference on NVIDIA GPUs using low-precision formats (FP8, FP4,...

GLIM is a factor-graph-based 3D mapping framework designed for multi-sensor point cloud localization and SLAM, supporting diverse range sensors...

CUTLASS is a specialized library providing CUDA C++ templates and Python DSLs for implementing high-performance matrix operations (GEMM) and linear...

UCCL is a GPU communication library optimized for collective operations, peer-to-peer transfers, and endpoint communication across heterogeneous GPU...

Beam is an open-source serverless runtime for AI workloads that emphasizes fast cold starts, GPU support, and developer experience through a Python...

WebGL Wind is a high-performance GPU-accelerated visualization library for rendering wind particle simulations, capable of displaying up to 1 million...

NVIDIA GPU Operator automates the provisioning and management of GPUs in Kubernetes clusters by orchestrating NVIDIA drivers, container runtimes,...

RAFT is a CUDA-accelerated C++ header-only library providing fundamental algorithms and primitives for machine learning, information retrieval, and...

cuML is a GPU-accelerated machine learning library that implements scikit-learn-compatible algorithms optimized for NVIDIA GPUs, delivering 10–50×...

MatX is a modern C++20 library for GPU-accelerated numerical computing on NVIDIA GPUs, offering Python-like syntax for operations like FFT, matrix...

Alacritty is a cross-platform, GPU-accelerated terminal emulator written in Rust, using OpenGL for rendering to achieve high throughput performance....