Evaluation of the Communication Motif for a Distributed Eigensolver using the SST Network Simulation Tool

A new motif that corresponds to the communication operations of the distributed LOBPCG eigensolver used in the Many-Fermion Dynamics–nuclear, or MFDn, code is constructed. The impact of communication strategy and process placement are evaluated on current and future architectures using the SST network simulation tool. Simulation of the communication motif is validated against production runs on the Cori system at NERSC. We identify the strengths and shortcomings of SST in doing so.

[1]  Andrew V. Knyazev,et al.  Toward the Optimal Preconditioned Eigensolver: Locally Optimal Block Preconditioned Conjugate Gradient Method , 2001, SIAM J. Sci. Comput..

[2]  Masha Sosonkina,et al.  Computational nuclear quantum many-body problem: The UNEDF project , 2013, Comput. Phys. Commun..

[3]  Joseph P. Kenny,et al.  Compiler-Assisted Source-to-Source Skeletonization of Application Models for System Simulation , 2018, ISC.

[4]  S. Liebig,et al.  Few-nucleon systems with state-of-the-art chiral nucleon-nucleon forces , 2015, 1505.07218.

[5]  Chao Yang,et al.  Accelerating nuclear configuration interaction calculations through a preconditioned block iterative eigensolver , 2016, Comput. Phys. Commun..

[6]  M. A. Caprio,et al.  Probing ab initio emergence of nuclear rotation , 2019 .

[7]  D. Roweth,et al.  Cray XC ® Series Network , 2012 .

[8]  C. Lanczos.  An iteration method for the solution of the eigenvalue problem of linear differential and integral operators , 1950 .

[9]  M. Sosonkina,et al.  N3LO NN interaction adjusted to light nuclei in ab exitu approach , 2016, 1605.00413.

[10]  Simon D. Hammond,et al.  (SAI) Stalled, Active and Idle: Characterizing Power and Performance of Large-Scale Dragonfly Networks , 2016, 2016 IEEE International Conference on Cluster Computing (CLUSTER).

[11]  J. Vary,et al.  Few- and many-nucleon systems with semilocal coordinate-space regularized chiral two- and three-body forces , 2018, Physical Review C.

[12]  Gilbert Hendry,et al.  SST: A Simulator for Exascale Co-design. , 2012 .

[13]  Nicholas J. Wright,et al.  GPCNeT: designing a benchmark suite for inducing and measuring contention in HPC networks , 2019, SC.

[14]  William J. Dally,et al.  Technology-Driven, Highly-Scalable Dragonfly Topology , 2008, 2008 International Symposium on Computer Architecture.

[15]  Chao Yang,et al.  Improving the scalability of a symmetric iterative eigensolver for multi‐core platforms , 2014, Concurr. Comput. Pract. Exp..

[16]  P. Navrátil,et al.  Origin of the Anomalous Long Lifetime of 14C , 2011, 1101.5124.

[17]  Taylor L. Groves,et al.  Simulation Framework for Studying Optical Cable Failures in Dragonfly Topologies , 2019, 2019 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW).

[18]  Petr Navratil,et al.  The Ab Initio No-core Shell Model , 2001, 0902.3510.

[19]  Samuel Williams,et al.  Optimizing Sparse Matrix-Multiple Vectors Multiplication for Nuclear Configuration Interaction Calculations , 2014, 2014 IEEE 28th International Parallel and Distributed Processing Symposium.

[20]  Masha Sosonkina,et al.  Scaling of ab-initio nuclear physics calculations on multicore computer architectures , 2010, ICCS.