# uccl-project/uccl

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)

Repository: https://github.com/uccl-project/uccl
Canonical: https://ross.abutalabs.com/products/uccl
Homepage: https://uccl-project.github.io/
Language: C++
License: Apache-2.0
License Family: permissive
Topics: ai, allreduce, amd, broadcom, collective, cuda, gpu, hpc, llm, networking, nvidia, rdma, kvcache, p2p, moe
Last push: 2026-08-23T03:58:17+00:00

## Health v2 (maintenance only)
Score: 71/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 51, longevity 43
- inputs: {"age_days": 604, "days_push": 10, "days_rel": 115, "gap_med": null, "n_releases_24m": 1}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1496, forks 169 (observed 2026-08-28T04:04:53.771143+00:00)

## What it is
UCCL is a high-performance GPU communication library written in C++ that provides collectives (as a drop-in NCCL/RCCL replacement), P2P transfers (KV cache, RL weight transfer), and expert-parallelism communication. It emphasizes flexibility for evolving ML workloads and portability across heterogeneous GPUs and network vendors (NVIDIA, AMD, Broadcom) using RDMA, InfiniBand, RoCE, and EFA.

## Use cases
- replace NCCL for faster AllReduce in multi-node LLM training
- transfer KV cache between GPUs for LLM inference disaggregation
- sync RL training weights across GPU nodes
- run expert-parallel MoE communication across NVIDIA and AMD GPUs
- communicate over RDMA, InfiniBand, RoCE, or AWS EFA networks
- speed up distributed training on cloud instances with commodity NICs

## When to choose
- you need faster collective communication than NCCL/RCCL provides
- your ML workloads span heterogeneous GPU vendors or network hardware
- you build LLM serving systems needing fast KV cache or weight transfer
- you run MoE models requiring efficient expert-parallel communication

## When to avoid
- you only need simple single-node GPU communication
- your stack depends strictly on NCCL-specific extensions
- you need a battle-tested, widely-deployed communication library rather than a newer research-driven one

## Facets
- artifact type: library
- maturity: active
- function: networking, gpu-computing, machine-learning, llm-training, llm-inference
- domain: machine-learning, gpu-computing, microservices, large-language-models, networking, performance
- platform: cpp, cross-platform
- tags: gpu-communication, nccl-replacement, rdma, allreduce, collective-communication, kv-cache-transfer, expert-parallelism, moe, rdma-transport, p2p-transfer, linux, gpu, cuda

## Member repositories
- uccl-project/uccl (main) score 71

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:53.771143+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:33:15.069875+00:00, confidence not recorded.
  - readme: https://github.com/uccl-project/uccl (fetched 2026-08-28T04:04:53.771143+00:00, sha 29497ed03b4d)
  - homepage: https://uccl-project.github.io/ (fetched 2026-08-29T11:38:30.394522+00:00, sha 02437c08d29f)
- Data as of 2026-08-30T08:39:29.467469+00:00.