NVIDIA Linux open GPU with P2P support

NCCL Profiling Kit

oneAPI Collective Communications Library (oneCCL)

One second to read GitHub code with VS Code.

Dissecting NVIDIA GPU Architecture

A paper list of spiking neural networks, including papers, codes, and related websites. 本仓库收集脉冲神经网络相关的顶会顶刊论文和代码，正在持续更新中。

torch_musa is an open source repository based on PyTorch, which can make full use of the super computing power of MooreThreads graphics cards.

A tool for bandwidth measurements on NVIDIA GPUs.

面向多平台编译优化的深度学习中间表示

LLM inference in C/C++

Mirage: Automatically Generating Fast GPU Kernels without Programming in Triton/CUDA

CUDA checkpoint and restore utility

[ASP-DAC 2025] "NeuronQuant: Accurate and Efficient Post-Training Quantization for Spiking Neural Networks" Official Implementation

Reinforcement learning environments for compiler and program optimization tasks

VPTQ, A Flexible and Extreme low-bit quantization algorithm

S-LoRA: Serving Thousands of Concurrent LoRA Adapters

AIOS: AI Agent Operating System

MSCCL++: A GPU-driven communication stack for scalable AI applications

A tool for examining GPU scheduling behavior.

Efficient Triton Kernels for LLM Training

Microsoft Azure Traces

Paella: Low-latency Model Serving with Virtualized GPU Scheduling

The Triton Inference Server provides an optimized cloud and edge inferencing solution.

Disaggregated serving system for Large Language Models (LLMs).

🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.

TiledCUDA is a highly efficient kernel template library designed to elevate CUDA C’s level of abstraction for processing tiles.

A fast communication-overlapping library for tensor parallelism on GPUs.

Scalable training and inference for Probabilistic Circuits