UCCL-Zip: Lossless Compression Supercharged GPU Communication

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ma, Shuang, Lao, Chon Lam, Xu, Zhiying, Wang, Zhuang, Mao, Ziming, Meng, Delong, Zhen, Jia, Wu, Jun, Stoica, Ion, Wang, Yida, Zhou, Yang
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910155931648000
author Ma, Shuang
Lao, Chon Lam
Xu, Zhiying
Wang, Zhuang
Mao, Ziming
Meng, Delong
Zhen, Jia
Wu, Jun
Stoica, Ion
Wang, Yida
Zhou, Yang
author_facet Ma, Shuang
Lao, Chon Lam
Xu, Zhiying
Wang, Zhuang
Mao, Ziming
Meng, Delong
Zhen, Jia
Wu, Jun
Stoica, Ion
Wang, Yida
Zhou, Yang
contents The rapid growth of large language models (LLMs) has made GPU communication a critical bottleneck. While prior work reduces communication volume via quantization or lossy compression, these approaches introduce numerical errors that can degrade convergence, accuracy, and stability. We present UCCL-Zip, a unified design that integrates lossless compression directly into GPU communication primitives. UCCL-Zip supports both point-to-point (P2P) and collective communication without modifying user-facing APIs or compromising numerical correctness. For P2P communication, Uzip-P2P employs a split-send pipeline that exposes transmissible data early and overlaps compression with communication, while preserving high GPU efficiency by operating on large data blocks. For collective communication, Uzip-NCCL integrates compression into NCCL's persistent kernel model via fused execution, eliminating redundant memory traffic and kernel launches. In real workloads, UCCL-Zip accelerates RL weight synchronization by up to 47.5% and reduces vLLM end-to-end inference latency by up to 10%, all without application changes.
format Preprint
id arxiv_https___arxiv_org_abs_2604_17172
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle UCCL-Zip: Lossless Compression Supercharged GPU Communication
Ma, Shuang
Lao, Chon Lam
Xu, Zhiying
Wang, Zhuang
Mao, Ziming
Meng, Delong
Zhen, Jia
Wu, Jun
Stoica, Ion
Wang, Yida
Zhou, Yang
Distributed, Parallel, and Cluster Computing
Artificial Intelligence
The rapid growth of large language models (LLMs) has made GPU communication a critical bottleneck. While prior work reduces communication volume via quantization or lossy compression, these approaches introduce numerical errors that can degrade convergence, accuracy, and stability. We present UCCL-Zip, a unified design that integrates lossless compression directly into GPU communication primitives. UCCL-Zip supports both point-to-point (P2P) and collective communication without modifying user-facing APIs or compromising numerical correctness. For P2P communication, Uzip-P2P employs a split-send pipeline that exposes transmissible data early and overlaps compression with communication, while preserving high GPU efficiency by operating on large data blocks. For collective communication, Uzip-NCCL integrates compression into NCCL's persistent kernel model via fused execution, eliminating redundant memory traffic and kernel launches. In real workloads, UCCL-Zip accelerates RL weight synchronization by up to 47.5% and reduces vLLM end-to-end inference latency by up to 10%, all without application changes.
title UCCL-Zip: Lossless Compression Supercharged GPU Communication
topic Distributed, Parallel, and Cluster Computing
Artificial Intelligence
url https://arxiv.org/abs/2604.17172