論文

査読有り
2018年7月

Practical Implementation of Lattice QCD Simulation on SIMD Machines with Intel AVX-512

LNCS 10962 (2018) 456-471
  • Issaku Kanamori
  • ,
  • Hideo Matsufuru

記述言語
英語
掲載種別
研究論文(国際会議プロシーディングス)
DOI
10.1007/978-3-319-95168-3_31

We investigate implementation of lattice Quantum Chromodynamics (QCD) code on<br />
the Intel AVX-512 architecture. The most time consuming part of the numerical<br />
simulations of lattice QCD is a solver of linear equation for a large sparse<br />
matrix that represents the strong interaction among quarks. To establish widely<br />
applicable prescriptions, we examine rather general methods for the SIMD<br />
architecture of AVX-512, such as using intrinsics and manual prefetching, for<br />
the matrix multiplication. Based on experience on the Oakforest-PACS system, a<br />
large scale cluster composed of Intel Xeon Phi Knights Landing, we discuss the<br />
performance tuning exploiting AVX-512 and code design on the SIMD architecture<br />
and massively parallel machines. We observe that the same code runs efficiently<br />
on an Intel Xeon Skylake-SP machine.

リンク情報
DOI
https://doi.org/10.1007/978-3-319-95168-3_31
arXiv
http://arxiv.org/abs/arXiv:1811.00893
URL
http://arxiv.org/abs/1811.00893v1