GPU acceleration of the Locally Selfconsistent Multiple Scattering code for first principles calculation of the ground state and statistical physics of materials

Markus Eisenbach; Jeff Larkin; Justin Lutjens; Steven Rennich; James H. Rogers

doi:10.1016/j.cpc.2016.07.013

GPU acceleration of the Locally Selfconsistent Multiple Scattering code for first principles calculation of the ground state and statistical physics of materials

Markus Eisenbach, Jeff Larkin, Justin Lutjens, Steven Rennich, James H. Rogers

NVIDIA

Research output: Contribution to journal › Article › peer-review

20 Scopus citations

Abstract

The Locally Self-consistent Multiple Scattering (LSMS) code solves the first principles Density Functional theory Kohn–Sham equation for a wide range of materials with a special focus on metals, alloys and metallic nano-structures. It has traditionally exhibited near perfect scalability on massively parallel high performance computer architectures. We present our efforts to exploit GPUs to accelerate the LSMS code to enable first principles calculations of O(100,000) atoms and statistical physics sampling of finite temperature properties. We reimplement the scattering matrix calculation for GPUs with a block matrix inversion algorithm that only uses accelerator memory. Using the Cray XK7 system Titan at the Oak Ridge Leadership Computing Facility we achieve a sustained performance of 14.5PFlop/s and a speedup of 8.6 compared to the CPU only code.

Original language	English
Pages (from-to)	2-7
Number of pages	6
Journal	Computer Physics Communications
Volume	211
DOIs	https://doi.org/10.1016/j.cpc.2016.07.013
State	Published - Feb 1 2017

Bibliographical note

Publisher Copyright:
© 2016 Elsevier B.V.

Funding

This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan ( http://energy.gov/downloads/doe-public-access-plan ). This work has been sponsored by the U.S. Department of Energy, Office of Science, Basic Energy Sciences, Material Sciences and Engineering Division (basic theory and applications) and by the Office of Advanced Scientific Computing (software optimization and performance measurements). This research used resources of the Oak Ridge Leadership Computing Facility, which is supported by the Office of Science of the U.S. Department of Energy under contract no. DE-AC05-00OR22725 .

Funders	Funder number
Office of Advanced Scientific Computing	DE-AC05-00OR22725
U.S. Department of Energy
Office of Science
Basic Energy Sciences

Keywords

First-principles
Monte-Carlo
Phase transitions

Access to Document

10.1016/j.cpc.2016.07.013

Cite this

@article{f03708e8b46c43178c078238205d0b20,

title = "GPU acceleration of the Locally Selfconsistent Multiple Scattering code for first principles calculation of the ground state and statistical physics of materials",

abstract = "The Locally Self-consistent Multiple Scattering (LSMS) code solves the first principles Density Functional theory Kohn–Sham equation for a wide range of materials with a special focus on metals, alloys and metallic nano-structures. It has traditionally exhibited near perfect scalability on massively parallel high performance computer architectures. We present our efforts to exploit GPUs to accelerate the LSMS code to enable first principles calculations of O(100,000) atoms and statistical physics sampling of finite temperature properties. We reimplement the scattering matrix calculation for GPUs with a block matrix inversion algorithm that only uses accelerator memory. Using the Cray XK7 system Titan at the Oak Ridge Leadership Computing Facility we achieve a sustained performance of 14.5PFlop/s and a speedup of 8.6 compared to the CPU only code.",

keywords = "First-principles, Monte-Carlo, Phase transitions",

author = "Markus Eisenbach and Jeff Larkin and Justin Lutjens and Steven Rennich and Rogers, {James H.}",

note = "Publisher Copyright: {\textcopyright} 2016 Elsevier B.V.",

year = "2017",

month = feb,

day = "1",

doi = "10.1016/j.cpc.2016.07.013",

language = "English",

volume = "211",

pages = "2--7",

journal = "Computer Physics Communications",

issn = "0010-4655",

publisher = "Elsevier",

}

TY - JOUR

T1 - GPU acceleration of the Locally Selfconsistent Multiple Scattering code for first principles calculation of the ground state and statistical physics of materials

AU - Eisenbach, Markus

AU - Larkin, Jeff

AU - Lutjens, Justin

AU - Rennich, Steven

AU - Rogers, James H.

PY - 2017/2/1

Y1 - 2017/2/1

N2 - The Locally Self-consistent Multiple Scattering (LSMS) code solves the first principles Density Functional theory Kohn–Sham equation for a wide range of materials with a special focus on metals, alloys and metallic nano-structures. It has traditionally exhibited near perfect scalability on massively parallel high performance computer architectures. We present our efforts to exploit GPUs to accelerate the LSMS code to enable first principles calculations of O(100,000) atoms and statistical physics sampling of finite temperature properties. We reimplement the scattering matrix calculation for GPUs with a block matrix inversion algorithm that only uses accelerator memory. Using the Cray XK7 system Titan at the Oak Ridge Leadership Computing Facility we achieve a sustained performance of 14.5PFlop/s and a speedup of 8.6 compared to the CPU only code.

AB - The Locally Self-consistent Multiple Scattering (LSMS) code solves the first principles Density Functional theory Kohn–Sham equation for a wide range of materials with a special focus on metals, alloys and metallic nano-structures. It has traditionally exhibited near perfect scalability on massively parallel high performance computer architectures. We present our efforts to exploit GPUs to accelerate the LSMS code to enable first principles calculations of O(100,000) atoms and statistical physics sampling of finite temperature properties. We reimplement the scattering matrix calculation for GPUs with a block matrix inversion algorithm that only uses accelerator memory. Using the Cray XK7 system Titan at the Oak Ridge Leadership Computing Facility we achieve a sustained performance of 14.5PFlop/s and a speedup of 8.6 compared to the CPU only code.

KW - First-principles

KW - Monte-Carlo

KW - Phase transitions

UR - http://www.scopus.com/inward/record.url?scp=84999837753&partnerID=8YFLogxK

U2 - 10.1016/j.cpc.2016.07.013

DO - 10.1016/j.cpc.2016.07.013

M3 - Article

AN - SCOPUS:84999837753

SN - 0010-4655

VL - 211

SP - 2

EP - 7

JO - Computer Physics Communications

JF - Computer Physics Communications

ER -

GPU acceleration of the Locally Selfconsistent Multiple Scattering code for first principles calculation of the ground state and statistical physics of materials

Abstract

Bibliographical note

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this