The Big Send-off: Scalable and Performant Collectives for Deep Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Singh, Siddharth, Pradeep, Keshav, Singh, Mahua, Wei, Cunyang, Bhatele, Abhinav
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915863493345280
author Singh, Siddharth
Pradeep, Keshav
Singh, Mahua
Wei, Cunyang
Bhatele, Abhinav
author_facet Singh, Siddharth
Pradeep, Keshav
Singh, Mahua
Wei, Cunyang
Bhatele, Abhinav
contents Collective communication is becoming increasingly important in data center and supercomputer workloads with an increase in distributed AI related jobs. However, existing libraries that provide collective support such as NCCL, RCCL, and Cray-MPICH exhibit several performance and scalability limitations on modern GPU supercomputers. To address these challenges, we introduce the Performant Collective Communication Library (PCCL), specifically targeted for distributed deep learning (DL) workloads. PCCL provides highly optimized implementations of key collectives used in distributed DL: all-gather, reduce-scatter, and all-reduce. PCCL uses a hierarchical design with learning-based adaptive selection of the best performing algorithms to scale efficiently to thousands of GPUs. It achieves substantial performance speedups over RCCL on 2048 GCDs of Frontier -- up to 168x for reduce-scatter, 33x for all-gather and 10x for all-reduce. More modest but still significant gains up to 5.7x over NCCL are observed on Perlmutter. These gains translate directly to performance improvement of production DL workloads: up to 4.9x speedup over RCCL in DeepSpeed ZeRO-3 training, and up to 2.4x speedup in DDP training.
format Preprint
id arxiv_https___arxiv_org_abs_2504_18658
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle The Big Send-off: Scalable and Performant Collectives for Deep Learning
Singh, Siddharth
Pradeep, Keshav
Singh, Mahua
Wei, Cunyang
Bhatele, Abhinav
Distributed, Parallel, and Cluster Computing
Artificial Intelligence
Machine Learning
Collective communication is becoming increasingly important in data center and supercomputer workloads with an increase in distributed AI related jobs. However, existing libraries that provide collective support such as NCCL, RCCL, and Cray-MPICH exhibit several performance and scalability limitations on modern GPU supercomputers. To address these challenges, we introduce the Performant Collective Communication Library (PCCL), specifically targeted for distributed deep learning (DL) workloads. PCCL provides highly optimized implementations of key collectives used in distributed DL: all-gather, reduce-scatter, and all-reduce. PCCL uses a hierarchical design with learning-based adaptive selection of the best performing algorithms to scale efficiently to thousands of GPUs. It achieves substantial performance speedups over RCCL on 2048 GCDs of Frontier -- up to 168x for reduce-scatter, 33x for all-gather and 10x for all-reduce. More modest but still significant gains up to 5.7x over NCCL are observed on Perlmutter. These gains translate directly to performance improvement of production DL workloads: up to 4.9x speedup over RCCL in DeepSpeed ZeRO-3 training, and up to 2.4x speedup in DDP training.
title The Big Send-off: Scalable and Performant Collectives for Deep Learning
topic Distributed, Parallel, and Cluster Computing
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2504.18658