102 PFLOPS lattice QCD quark solver on Fugaku - INSPIRE

DataBETA

102 PFLOPS lattice QCD quark solver on Fugaku

Ken-Ichi Ishikawa
(
- Hiroshima U. (main)
)
,
Issaku Kanamori
(
- RIKEN AICS, Kobe
)
,
Hideo Matsufuru
(
- KEK, Tsukuba
)
,
Ikuo Miyoshi
(
- Fijitsu, Tsukuba
)
,
Yuta Mukai
(
- Fijitsu, Tsukuba
)

Sep 22, 2021

27 pages

Published in:

Comput.Phys.Commun. 282 (2023) 108510

Published: Jan, 2023

e-Print:

2109.10687 [hep-lat]

DOI:

10.1016/j.cpc.2022.108510 (publication)

Report number:

HUPD-2108

View in:

ADS Abstract Service

reference search14 citations

Citations per year

Abstract: (Elsevier B.V.)

We present results on the world's first over 100 PFLOPS single precision lattice QCD quark solver on the Japanese new supercomputer Fugaku. We achieve a factor 38 time speedup from the supercomputer K on the same problem size, 192⁴, with 102 PFLOPS, 10% floating-point operation efficiency against single precision floating-point operation peak. The evaluation region is the single precision BiCGStab for a Clover–Wilson Dirac matrix with Schwarz Alternating Procedure domain decomposition preconditioning using Jacobi iteration for the local domain matrix inversion.

Note:

27 pages, 8 figures

High performance computing
Lattice QCD
Iterative solver
A64FX
lattice field theory
quark
efficiency

References(19)

Figures(9)

[1]

R.F. Barrett

- Transition of HPC Towards Exascale Computing vol. 24 (2013) 141-155

[2]

M. Sato

[3]

TOP500

- https://www.top500.org

[4]

A64FX, Arm microprocessor which conforms to Armv8 + SVE

- https://github.com/fujitsu/A64FX

[5]

Y. Ajima

[6]

Y. Nakamura
,
Y. Mukai
,
K.-I. Ishikawa
,
I. Kanamori

- https://github.com/RIKEN-LQCD/qws

[7]

Multigrid solver on Fugaku

Ken-Ichi Ishikawa
(
- Hiroshima U.
)
,
Issaku Kanamori
(
- RIKEN (main)
)
,
Hideo Matsufuru
(
- KEK, Tsukuba and
- Sokendai, Tsukuba
)

- PoS LATTICE2021 (2022) 278
•
e-Print:
- 2112.00501
•
DOI:
- 10.22323/1.396.0278

[8]

M. Lüscher

- J. High Energy Phys. 0305 (2003) 119

[9]

A. Buttari

- ACM Trans. Math. Softw. 34 (2008) 1
•
DOI:
- 10.1145/1377596.1377597

[10]

Y. Saad

- SIAM J.Sci.Comput. 14 (1993) 2, 461

[11]

J.A. Vogel

- Appl.Math.Comput. 188 (2007) 1, 226

[12]

T. Boku

[13]

Solution of the Dirac equation in lattice QCD using a domain decomposition method

Martin Luscher
(
- CERN
)

- Comput.Phys.Commun. 156 (2004) 209-220
•
e-Print:
- hep-lat/0310048
•
DOI:
- 10.1016/S0010-4655(03)00486-7

[14]

H. Tadano
,
T. Sakurai

- Lec. Notes Comput. Sci. vol. 4818 (2008) 721

[15]

Arm C Language Extensions

- https://developer.arm.com/architectures/system-architectures/software-standards/acle

[16]

Y. Ajima
,
S. Sumimoto
,
T. Shimizu

- IEEE Comput. Soc. 42 (2009) 11, 36-40
•
DOI:
- 10.1109/MC.2009.370

[17]

P. Boyle

- Nucl.Phys.B Proc.Suppl. 169-175

[18]

K.-I. Ishikawa

[19]

McKernel

- https://github.com/RIKEN-SysSoft/mckernel