Wilson and Domainwall Kernels on Oakforest-PACS

We report the performance of Wilson and Domainwall Kernels on a new Intel Xeon Phi Knights Landing based machine named Oakforest-PACS, which is co-hosted by University of Tokyo and Tsukuba University and is currently fastest in Japan. This machine uses Intel Omni-Path for the internode network. We compare performance with several types of implementation including that makes use of the Grid library. The code is incorporated with the code set Bridge++.

Note:

8 pages, 9 figures, Proceedings for the 35th International Symposium on Lattice Field Theory (Lattice 2017)

numerical methods: efficiency
quantum chromodynamics
lattice field theory
numerical calculations
microprocessor
computer: performance
Grid computing
network
fermion: Wilson
fermion: domain wall

References(10)

Figures(9)

Bridge++ project

Lattice QCD on new chips: a community summary, in Proceedings, 35th International Symposium on Lattice Field Theory (Lattice): Granada, Spain, to appear in EPJWeb Conf. [Google Scholar]

A. Rago

Joint Center for Advanced High Performance Computing (JCAHPC)

Intel Xeon Phi Processor High Performance Programming Knights Landing Edition

A.S.J. Jeffers
,
J. Reinders

Getting the most out of Intel(R) Xeon Phi(tm) processor, in Proceedings, 35th International Symposium on Lattice Field Theory (Lattice): Granada, Spain, to appear in EPJ Web Conf. [Google Scholar]

H. Servat

Lattice QCD code set Bridge++ on arithmetic accelerators

et al.

- PoS LATTICE2015 (2016) 040
•
DOI:
- 10.22323/1.251.0040

(Bridge++ Project), Procedia Computer Scienc 51, 1313 [Google Scholar]

H. Matsufuru
,
S.Aoki
,
T.Aoyama
,
K.Kanaya
,
S.Motoki

et al.

•
DOI:
- 10.1016/j.procs.2015.05.316

Grid: A next generation data parallel C++ QCD library

Peter A. Boyle
(
- Edinburgh U.
)
,
Guido Cossu
(
- INFN, Pisa and
- Pisa, Scuola Normale Superiore
)
,
Azusa Yamaguchi
(
- Glasgow U.
)
,
Antonin Portelli
(
- Southampton U.
)

- PoS LATTICE2015 (2016) 023
•
DOI:
- 10.22323/1.251.0023

[Google Scholar]

to appear in Proceedings, 5th International Workshop on Legacy HPC Application Migration (LHAM’17) in The fifth International Symposium on Computing and Networking (CANDAR’17): Aomori, Japan [Google Scholar]

I. Kanamori
,
H. Matsufuru