PipeDream: Fast and Efficient Pipeline Parallel DNN Training

Harlap, Aaron; Narayanan, Deepak; Phanishayee, Amar; Seshadri, Vivek; Devanur, Nikhil; Ganger, Greg; Gibbons, Phil

51 captures

12 Dec 2019 - 03 Mar 2026

Apr	MAY	Jun
	17
2022	2023	2024

success

fail

About this capture

COLLECTED BY

Collection: Save Page Now Outlinks

TIMESTAMPS

The Wayback Machine - http://web.archive.org/web/20230517232327/https://arxiv.org/abs/1806.03377

We gratefully acknowledge support from the Simons Foundation and member institutions.

> cs> arXiv:1806.03377

quick links

●Login ●Help Pages ●About

Computer Science > Distributed, Parallel, and Cluster Computing

arXiv:1806.03377 (cs) [Submitted on 8 Jun 2018]

Title:PipeDream: Fast and Efficient Pipeline Parallel DNN Training

Authors:Aaron Harlap, Deepak Narayanan, Amar Phanishayee, Vivek Seshadri, Nikhil Devanur, Greg Ganger, Phil Gibbons Download a PDF of the paper titled PipeDream: Fast and Efficient Pipeline Parallel DNN Training, by Aaron Harlap and 6 other authors Download PDF Abstract: PipeDream is a Deep Neural Network(DNN) training system for GPUs that parallelizes computation by pipelining execution across multiple machines. Its pipeline parallel computing model avoids the slowdowns faced by data-parallel training when large models and/or limited network bandwidth induce high communication-to-computation ratios. PipeDream reduces communication by up to 95% for large DNNs relative to data-parallel training, and allows perfect overlap of communication and computation. PipeDream keeps all available GPUs productive by systematically partitioning DNN layers among them to balance work and minimize communication, versions model parameters for backward pass correctness, and schedules the forward and backward passes of different inputs in round-robin fashion to optimize "time to target accuracy". Experiments with five different DNNs on two different clusters show that PipeDream is up to 5x faster in time-to-accuracy compared to data-parallel training.

Subjects:	Distributed, Parallel, and Cluster Computing (cs.DC)
Cite as:	arXiv:1806.03377 [cs.DC]
	(or arXiv:1806.03377v1 [cs.DC] for this version)
	https://doi.org/10.48550/arXiv.1806.03377

Submission history

From: Deepak Narayanan [view email] [v1] Fri, 8 Jun 2018 23:18:08 UTC (891 KB) Full-text links:

Download:

Download a PDF of the paper titled PipeDream: Fast and Efficient Pipeline Parallel DNN Training, by Aaron Harlap and 6 other authors PDF ●Other formats (license) Current browse context: cs.DC < prev | next > new | recent | 1806 Change to browse by: cs

BibTeX formatted citation

Data provided by:

Bookmark

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer (What is the Explorer?) Litmaps (What is Litmaps?) scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

DagsHub (What is DagsHub?) Papers with Code (What is Papers with Code?) ScienceCast (What is ScienceCast?)

Demos

Replicate (What is Replicate?) Hugging Face Spaces (What is Spaces?) Related Papers

Recommenders and Search Tools

Influence Flower (What are Influence Flowers?) Connected Papers (What is Connected Papers?) CORE Recommender (What is CORE?)

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) ●About ●Help ● ntact arXiv Contact ● ick here to subscribe Subscribe ●Copyright ●Privacy Policy ●Web Accessibility Assistance Get status notifications via emailorslack