Skip to content
@uccl-project

UCCL

Next-generation GPU communication

Pinned Loading

  1. uccl uccl Public

    UCCL is an efficient communication library for GPUs, covering collectives, P2P (e.g., KV cache transfer, RL weight transfer), and EP (e.g., GPU-driven)

    C++ 1.5k 179

  2. mKernel mKernel Public

    mKernel: fast multi-node, multi-GPU fused kernels

    Cuda 292 29

  3. CommBench CommBench Public

    Can LLMs Write Correct and Efficient GPU Communication Code?

    Python 66 2

  4. rdmatop rdmatop Public

    htop-like TUI for real-time RDMA network monitoring.

    Rust 114 10

Repositories

Showing 9 of 9 repositories
  • mKernel Public

    mKernel: fast multi-node, multi-GPU fused kernels

    uccl-project/mKernel's past year of commit activity
    Cuda 292 MIT 29 2 5 Updated Oct 11, 2026
  • uccl Public

    UCCL is an efficient communication library for GPUs, covering collectives, P2P (e.g., KV cache transfer, RL weight transfer), and EP (e.g., GPU-driven)

    uccl-project/uccl's past year of commit activity
    C++ 1,540 Apache-2.0 179 57 (1 issue needs help) 12 Updated Oct 10, 2026
  • rdmatop Public

    htop-like TUI for real-time RDMA network monitoring.

    uccl-project/rdmatop's past year of commit activity
    Rust 114 Apache-2.0 10 0 0 Updated Oct 2, 2026
  • CommBench Public

    Can LLMs Write Correct and Efficient GPU Communication Code?

    uccl-project/CommBench's past year of commit activity
    Python 66 2 0 1 Updated Jul 7, 2026
  • uccl-project.github.io Public
    uccl-project/uccl-project.github.io's past year of commit activity
    Astro 3 MIT 3 1 1 Updated Jun 14, 2026
  • nixl Public Forked from ai-dynamo/nixl

    NVIDIA Inference Xfer Library (NIXL)

    uccl-project/nixl's past year of commit activity
    C++ 0 Apache-2.0 479 0 0 Updated Nov 24, 2025
  • vllm Public Forked from vllm-project/vllm

    A high-throughput and memory-efficient inference and serving engine for LLMs

    uccl-project/vllm's past year of commit activity
    Python 0 Apache-2.0 23,535 0 1 Updated Nov 3, 2025
  • ray-uccl Public Forked from ray-project/ray

    Ray is an AI compute engine. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.

    uccl-project/ray-uccl's past year of commit activity
    Python 0 Apache-2.0 8,330 0 0 Updated Jul 22, 2025
  • nccl Public Forked from NVIDIA/nccl

    Optimized primitives for collective multi-GPU communication

    uccl-project/nccl's past year of commit activity
    C++ 0 1,454 0 0 Updated Jul 4, 2025