Mixed Precision Solver Scalable to 16000 MPI Processes for Lattice Quantum Chromodynamics Simulations on the Oakforest-PACS System
نویسندگان
چکیده
Taisuke Boku∗†, Ken-Ichi Ishikawa‡§ , Yoshinobu Kuramashi¶† and Lawrence Meadows‖ ∗Graduate School of Systems and Information Engineering, University of Tsukuba, Tsukuba, Ibaraki 305-8573, Japan †Center for Computational Sciences, University of Tsukuba, Tsukuba, Ibaraki 305-8577, Japan ‡Graduate School of Science, Hiroshima University, Higashi-Hiroshima, Hiroshima 739-8526, Japan §Core of Research for the Energetic Universe, Hiroshima University, Higashi-Hiroshima, Hiroshima 739-8526, Japan ¶Graduate School of Pure and Applied Sciences, University of Tsukuba, Tsukuba, Ibaraki 305-8571, Japan ‖Intel Corporation, USA
منابع مشابه
Nuclear Force from Monte Carlo Simulations of Lattice Quantum Chromodynamics
The nuclear force acting between protons and neutrons is studied in the Monte Carlo simulations of the fundamental theory of the strong interaction, the quantum chromodynamics defined on the hypercubic space-time lattice. After a brief summary of the empirical nucleon-nucleon (NN) potentials which can fit the NN scattering experiments in high precision, we outline the basic formulation to deriv...
متن کاملA Scalable Parallel SSOR Preconditioner for Efficient Lattice Computations in Gauge Theories
solver We extend the parallel SSOR procedure for the eecient preconditioning of modern Krylov subspace solvers 1], recently introduced in 2] towards higher order, quantum-improved discretization schemes 3] for lattice quantum chromodynamics 4]. 1. Solving linear systems in LGT. Lattice gauge theory (LGT) deals with the controlled numerical evaluation of gauge theories like quantum chromodynamic...
متن کاملScalable Parallel SSOR Preconditioning for Lattice Computations in Gauce Theories
We discuss a parallelization scheme for SSOR precondition-ing of Krylov subspace solvers as applied in lattice gauge theory computations. Our preconditioner is based on a locally lexicographic ordering of the lattice points leading to a parallelism adapted to the parallel system's size. By exploitation of thèEisenstat-trick' within the bi-conjugate gradient stabilized iterative solver, we achie...
متن کاملExtreme Scaling of Lattice Quantum Chromodynamics
As the complexity and size of challenges in science and engineering are continually increasing, it is highly important that applications are able to scale strongly to very large numbers of cores (>100,000 cores) to enable HPC systems to be utilised efficiently. This paper presents results of strong scaling tests performed with an MPI only and a hybrid MPI + OpenMP version of the Lattice QCD app...
متن کاملA fully distributed CFD framework for massively parallel systems
A highly scalable solver framework, based on a linearized octree is presented. It allows for fully distributed computations and avoids special needs on single processes with potential bottlenecks, while enabling simulations with complex geometries. Scaling results on the Cray XE6 system Hermit at HLRS in Stuttgart are presented with runs up to 3072 nodes with 98304 MPI processes. Even with a fu...
متن کامل