Benchmark Distribution & Run Rules
Applications and micro-benchmarks for the Crossroads/NERSC-9 procurement. You can find more information by clicking on the header for each of the topics listed below.
Change and update notes for the benchmark suite.
The following applications will be used by the Sustained System Improvement metric in measuring the performance improvement of proposed systems relative to NERSC's Edison platform.
Application specific run rules supersede this description. Anything unclear should be reported to the APEX procurement team.
Description and definition of the Scalable System Improvement metric. The spreadsheet called out in the Technical Requirements document for the calculation of SSI can be found here. The table below provides the reference values obtained on Edison. More detail can be found in the spreadsheet. The values in the table are tentative and are subject to change until the final Request for Proposals (RFP) is issued.
|Application||# of Nodes (n)
||Time or FOM
||# MPI ranks||# OMP threads|
|Meraculous||400||42.74 secs||9600 (UPC)||N/A|
|MiniPIC||1024||1.12 E9 updates/sec||4096||6|
|SNAP||2592||1.3659 E-3 secs||10368||6|
3D Gyrokinetic Toroidal code.
High Performance Conjugate Gradient benchmark.
A massively parallel de novo whole genome assembly benchmark.
MILC represents part of a set of codes used to study Lattice Quantum Chromodynamics (QCD).
MiniDFT is intended to capture the performance-critical portions of a density functional theory materials science computation.
Particle in cell code that solves the discrete Boltzman equation in an electrostatic field in an arbitrary domain with reflective walls.
A mini-application for 2D, unstructured, finite element mesh with arbitrary polygons.
SNAP is a proxy for the performance of a modern discrete ordinates neutral particle transport application.
UMT is a proxy application (mini-app) that performs three-dimensional, non-linear, radiation transport calculations using deterministic (Sn) methods.
The following microbenchmarks will be used in support of specific requirements in the RFP.
The DGEMM benchmark measures the sustained floating-point rate of a single node.
IOR is used for testing performance of parallel file systems using various interfaces and access patterns.
A metadata benchmark that performs open/stat/close operations on files and directories.
The STREAM benchmark measures sustainable memory bandwidth using four simple vector kernels.