Parallelizing the QUDA library for multi-GPU calculations in lattice quantum chromodynamics

Ronald Babich, Michael A. Clark, Bálint Joó

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

70 Scopus citations

Abstract

Graphics Processing Units (GPUs) are having a transformational effect on numerical lattice quantum chromo-dynamics (LQCD) calculations of importance in nuclear and particle physics. The QUDA library provides a package of mixed precision sparse matrix linear solvers for LQCD applications, supporting single GPUs based on NVIDIA's Compute Unified Device Architecture (CUDA). This library, interfaced to the QDP++/Chroma framework for LQCD calculations, is currently in production use on the"9g" cluster at the Jefferson Laboratory, enabling unprecedented price/performance for a range of problems in LQCD. Nevertheless, memory constraints on current GPU devices limit the problem sizes that can be tackled. In this contribution we describe the parallelization of the QUDA library onto multiple GPUs using MPI, including strategies for the overlapping of communication and computation. We report on both weak and strong scaling for up to 32 GPUs interconnected by InfiniBand, on which we sustain in excess of 4 Tflops.

Original languageEnglish
Title of host publication2010 ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2010
DOIs
StatePublished - 2010
Externally publishedYes
Event2010 ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2010 - New Orleans, LA, United States
Duration: Nov 13 2010Nov 19 2010

Publication series

Name2010 ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2010

Conference

Conference2010 ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2010
Country/TerritoryUnited States
CityNew Orleans, LA
Period11/13/1011/19/10

Funding

FundersFunder number
National Science Foundation0946441, 0835713

    Fingerprint

    Dive into the research topics of 'Parallelizing the QUDA library for multi-GPU calculations in lattice quantum chromodynamics'. Together they form a unique fingerprint.

    Cite this