LP Benchmark
Benchmark Guidelines
This site contains the benchmarks required for the 2017 University of Toronto Large Parallel System procurement. The benchmark sources and specific run instructions are provided for each of the benchmarks below. Information on baseline, optimization and reporting of the benchmark scores is described in Schedule 1 of the procurement document.
Large Parallel Benchmark (LPBM)
The following six (6) application benchmarks make up the Large Parallel Benchmark (LPBM). Each benchmark includes the source codes, benchmark run requirements and the instructions for reporting.
HPCG
The High Performance Conjugate Gradients HPCG benchmark is designed to exercise computational and data access patterns that closely match a broad set of important scientific applications, and to give incentive to computer system designers to invest in capabilities that will have impact on the collective performance of these applications.
HPCG is an open source benchmark developed by Mike Heroux, Jack Dongarra and Piotr Luszcze
Source and LPBM specific instructions can be downloaded here
Configure and build instructions for Power here
Nek5000
Nek5000 is an open source, highly scalable and portable spectral element code designed to simulate a range of flow physics.
Developed at Argonne National Laboratory, Nek5000 is released under the terms of the GNU GPL license
Source and LPBM specific instructions can be downloaded here
Configure and build instructions for Power here
WRF
The Weather Research and Forecasting (WRF) Model is a next-generation mesoscale numerical weather prediction system designed for both atmospheric research and operational forecasting needs.
WRF is public domain, details here
Source can be downloaded here
Configure and build instructions for Power here
Input files and LPBM specific instructions can be downloaded here (NOTE: 36G file) MD5SUMS here
NAMD
NAMD is a parallel molecular dynamics code designed for high-performance simulation of large biomolecular systems.
NAMD is Licensed under the University of Illinois NAMD Molecular Dynamics Software Non-Exclusive, Non-Commercial Use License http://www.ks.uiuc.edu/Research/namd/license.html
NAMD version 2.12 source, and/or binary packages, must be obtained directly from the Theoretical and Computational Biophysics Group at UIUC here.
The benchmark to be used is the simulation of the "Satellite Tobacco Mosaic Virus" (see here).
LPBM specific instructions and input files can be downloaded here.
miniDFT
MiniDFT is a plane-wave density functional theory (DFT) mini-app for modeling materials. Given a set of atomic coordinates and pseudopotentials, MiniDFT computes self-consistent solutions of the Kohn-Sham equations using either the LDA or PBE exchange-correlation functionals. For each iteration of the self-consistent field cycle, the Fock matrix is constructed and then diagonalized. To build the Fock matrix, Fast Fourier Transforms are used to transform orbitals from the plane wave basis ( where the kinetic energy is most readily computed ) to real space (where the potential is evaluated ) and back. Davidson diagonalization is used to compute the orbital energies and update the orbital coefficients.
The MiniDFT mini-app was excised from the general-purpose Quantum Espresso (QE) code. Quantum Espresso is licensed per the GNU General Public License (GPL). A copy of the GPL is provided in the distribution's 'License' file.
Source and LPBM specific instructions can be downloaded here
SPEC MPI 2007
SPEC MPI® 2007 focuses on performance of compute intensive applications using the Message-Passing Interface (MPI)library. All submissions are expected to follow the run rules, found online at https://www.spec.org/mpi2007/Docs/runrules.html#section1.1.4 . Any deviations from run rules should be described fully in an accompanying document."
Proponents should report the SPECmpiL_base2017 and optionally the SPECmpiL_peak2017 scores in the Schedule 1 LPBM spreadsheet for the closest number of whole nodes to achieve 2048 MPI ranks. All input and output should be submitted as well.
Available from https://www.spec.org/mpi2007/
IOR
IOR is designed to measure parallel file system I/O performance at both the POSIX and MPI-IO level. "IOR" stands "InterleavedOrRandom, which has very little to do with how the program works currently. This parallel program performs writes and reads to/from files under several sets of conditions and reports the resulting throughput rates.
Source and specific instructions can be downloaded here
Benchmark Output Upload
Upload the requested benchmark outputs, one tar.gz file for each benchmark, to https://niagara.scinet.utoronto.ca/upload.html.
Problems
If you encounter problems downloading the benchmarks or uploading the benchmark output, please email support@scinet.utoronto.ca.