HPC Advisory Council Best Practices

The HPC Advisory Council provides best practices, that through experience and research, have shown to improve clustering and applications productivity. The HPC Advisory Council is encouraging HPC users to provide their feedback and suggest other application or HPC areas as candidates for future best practices.

Cases: OpenAtom

OpenAtom is a highly scalable and portable parallel application for molecular dynamics simulations at the quantum level. It implements the Car-Parrinello ab-initio Molecular Dynamics (CPAIMD) method. OpenAtom is written using the Charm++ parallel programming framework. It runs on a variety of architectures like PowerPC, Opteron and Intel-based systems.  The presentation provides information on OpenAtom productivity, how to optimize and compile the code for highest performance and efficiency, and the effect HPC cluster components (HW and SW) have on OpenAtom performance.

Cases: OpenFOAM

OpenFOAM (Open Field Operation and Manipulation) CFD Toolbox is a free, open source CFD software package produced by a commercial company, OpenCFD Ltd. It has a large user base across most areas of engineering and science, from both commercial and academic organizations. OpenFOAM has an extensive range of features to solve anything from complex fluid flows involving chemical reactions, turbulence and heat transfer, to solid dynamics and electromagnetics. The presentation provides information on OpenFOAM performance, how to optimize and compile the code for highest performance and efficiency, and the effect HPC cluster components (HW and SW) have on OpenFOAM performance.

Cases: Abaqus

The Abaqus software suite consists of three core products: Abaqus/Standard, Abaqus/Explicit and Abaqus/CAE. Abaqus/Standard is a general-purpose solver using a traditional implicit integration scheme to solve finite element analyses. Abaqus/Explicit uses an explicit integration scheme to solve highly nonlinear transient dynamic and quasi-static analyses. Abaqus/CAE provides an integrated modeling (preprocessing) and visualization (post-processing) environment for the analysis products. Abaqus is used in the automotive, aerospace, and industrial product industries. The product is popular with academic and research institutions due to the wide material modeling capability, and the program's ability to be customized. Abaqus also provides a good collection of multiphysics capabilities, such as coupled acoustic-structural, piezoelectric, and structural-pore capabilities, making it attractive for production-level simulations where multiple fields need to be coupled. The presentation provides information on Abaqus performance capabilities and the effect that different HPC cluster components (HW and SW) have on it.

Cases:  Dacapo

Dacapo is a total energy program based on density functional theory that uses a plane wave basis for the valence electronic states. It describes core-electron interactions with Vanderbilt ultrasoft pseudo-potentials and performs molecular dynamics / structural relaxation simultaneous. Dacapo is an open-source code, maintained by Technical University of Denmark. The presentation provides information on Dacapo performance capabilities and the effect that different HPC cluster components (HW and SW) have on it.

Cases:  LAMMPS

LAMMPS (Large-scale Atomic/Molecular Massively Parallel Simulator) has potentials for soft materials (biomolecules, polymers) and solid-state materials (metals, semiconductors) and coarse-grained or mesoscopic systems. It can be used to model atoms or, more generically, as a parallel particle simulator at the atomic, meso, or continuum scale. LAMMPS is distributed as an open source code under the terms of the GPL. LAMMPS is distributed by Sandia National Laboratories, a US Department of Energy laboratory. Funding for LAMMPS development has come primarily from DOE (OASCR, OBER, ASCI, LDRD, Genomes-to-Life). The presentation provides information on LAMMPS performance capabilities and the effect that different HPC cluster components (HW and SW) have on it.

Cases:  Desmond

Desmond is a software package developed at D. E. Shaw Research to perform high-speed molecular dynamics simulations of biological systems on conventional commodity clusters. The code uses novel parallel algorithms and numerical techniques to achieve high-performance and accuracy on platforms containing a large number of processors, but may also be executed on a single computer. The presentation provides information on Desmond performance capabilities and the effect that different HPC cluster components (HW and SW) have on it.

Cases:  ANSYS CFX

ANSYS CFX is a high-performance, general purpose CFD program that has been applied to solve wide-ranging fluid flow problems. At the heart of ANSYS CFX is its advanced solver technology, the key to achieving reliable and accurate solutions quickly and robustly. The highly parallelized solver is the foundation for an abundant choice of physical models to capture virtually any type of phenomena related to fluid flow. The presentation provides information on CFX performance capabilities and the effect of the different HPC cluster components (HW and SW) has on it.

Cases: NWChem

NWChem is a computational chemistry package developed by the Molecular Sciences Software group of the Environmental Molecular Sciences Laboratory (EMSL) at the Pacific Northwest National Laboratory (PNNL). NWChem provides many methods to compute the properties of molecular and periodic systems using standard quantum mechanical descriptions of the electronic wavefunction or density. NWChem has the capability to perform classical molecular dynamics and free energy simulations and these approaches may be combined to perform mixed quantum-mechanics and molecular-mechanics simulations. The presentation includes performance data, comparisons between mathematical libraries, and power/performance results.

Cases: The Massively Parallel Quantum Chemistry Program (MPQC)

MPQC computes properties of atoms and molecules from first principles using the time independent Schrödinger equation. The presentation include performance comparisons between different interconnect technologies, productivity results, power aware simulations and performance results with CPU frequency scaling.

Cases: ANSYS FLUENT

FLUENT is a leading CFD application from ANSYS that is being used for solving fluid flow problems. The broad physical modeling capabilities of FLUENT have been applied to industrial applications ranging from air flow over an aircraft wing to combustion in a furnace, from bubble columns to glass production, from blood flow to semiconductor manufacturing, from clean room design to wastewater treatment plants. The presentation provides information FLUENT performance capabilities and the effect different HPC cluster components (HW and SW) has on it, a comparison between FLUENT 12 and 6.3 and a power aware and productivity aware usage model.

Cases: STAR-CCM+ and STAR-CD

CD-adapco is a leading global provider of full-spectrum engineering simulation (CAE) solutions for fluid flow, heat transfer and stress. CD-adapco core products are the technology-leading simulation packages, STAR-CCM+ and STAR-CD. STAR-CCM+ is an engineering process oriented Computational Fluid Dynamics tool that delivers the latest CFD technology in a single integrated environment. STAR-CD is an integrated platform for performing powerful multi-physics simulations, unrivalled in its ability to tackle problems involving multi-physics and complex geometries. The presentation provides information on both STAT-CD and STAR-CCM+ performance capabilities and the effect different HPC cluster components (HW and SW) has on it, as well as a power aware and productivity aware usage model.

Cases: The Fifth-Generation Mesoscale Model (MM5)

The Fifth-Generation NCAR / Penn State Mesoscale Model (MM5) is a limited-area, nonhydrostatic or hydrostatic, terrain-following sigma-coordinate model designed to simulate or predict mesoscale and regional-scale atmospheric circulation. It has been developed at Penn State and NCAR as a community mesoscale model. Mesoscale Meteorology is the study of weather systems smaller than synoptic scale systems but larger than microscale and storm-scale cumulus systems (horizontal dimensions generally range from around 5 kilometers to several hundred kilometers).

The presentation provides information on MM5 performance capabilities and the effect of different HPC cluster components (HW and SW) on it, as well and power aware usage model.

Cases: Molecular Dynamics (NAMD)

NAMD is a parallel molecular dynamics code designed for high-performance simulations of large biomolecular systems and scales to hundreds of processors on high-end parallel platforms. NAMD was developed by the joint collaboration of the Theoretical and Computational Biophysics Group (TCB) and the Parallel Programming Laboratory (PPL) at the University of Illinois at Urbana-Champaign, and is distributed free of charge with source code. The presentation provides information on NAMD performance capabilities and the effect of different HPC cluster components (HW and SW) on NAMD.

Cases: Automotive Crash Simulation (LS-DYNA)

One of the most demanding applications of automotive design is crash simulation (full-frontal, offset-frontal, angle-frontal, side-impact, rear-impact and more). Crash simulations, while performed very early in the development process, are validated very late in the development process once the vehicle is completely built. The more sophisticated and complex the simulation, the more parts and details can be analyzed. Automotive makers increase their dependency for car crash simulations throughout the design process while reducing the need for real prototypes, thus achieving faster time to market with less cost associated with the design phase.

LS-DYNA is a general purpose structural and fluid analysis simulation software package capable of simulating complex real world problems. It is widely used in the automotive industry for crashworthiness, occupant safety and metal forming and also for aerospace, military and defense and consumer products. The presentation provides a deep analysis on LS-DYNA performance and scalability on HPC clusters, and provide recommendations for improving it’s productivity.

Cases: Oil and Gas Reservoir Simulation (Eclipse)

Reservoir simulation is a core technology used for most of the decisions undertaken in the upstream oil industry to predict plateau levels of fields, calculate the number of wells to be drilled, select well locations, estimate facility requirements, calculate reserves depletions, and design reservoir management strategies for recovering more oil and gas.

Schlumberger ECLIPSE reservoir simulation is one of the most used software solutions that allows engineers to predict and manage fluid flow more efficiently, with greater insight and better accurate modeling. The presentation provides recommendations for improving Eclipse performance, scalability, and productivity as measured in jobs per day, by exploring and profiling the software on HPC clusters.

Cases: Car-Parrinello Molecular Dynamics (CPMD)

Car-Parrinello Molecular Dynamics (CPMD) is an ab initio electronic structure and molecular dynamics (MD) simulation software that provides a powerful way to perform molecular dynamic simulations from first principles, using a plane wave/pseudopotential implementation of density functional theory. The CPMD code has been used to examine systems including protein active sites, liquid-surface interactions, and surface catalysts. The ability to examine interactions on the nanoscale makes this approach ideal for studying systems where chemical and biological interactions are critical. The presentations provides recommendations for improving CPMD performance, scalability, and productivity as measured in jobs per day.

Cases: The Weather Research and Forecast (WRF) Model

The Weather Research and Forecast (WRF) Model is a fully functioning modeling system for atmospheric research and operational weather prediction communities. With an emphasis on efficiency, portability, maintainability, scalability and productivity, WRF has been successfully deployed over the years on a wide variety of HPC clustered compute nodes connected with high speed interconnects - the most used system architecture for high-performance computing. As such, understanding WRF dependency on the various clustering elements, such as the CPU, interconnects and the software libraries are crucial for enabling efficient predictions and high productivity. Our results identify WRF’s communication-sensitive points and demonstrate WRF’s dependency on high-speed networks and fast CPU-to-CPU communication. Both factors are critical to maintaining scalability and increasing productivity when adding cluster nodes. We conclude with specific recommendations for improving WRF performance, scalability, and productivity as measured in jobs per day. Because proprietary hardware and software can quickly erode cluster architecture’s favorable economics, we have restricted our investigation to standards based hardware and open source software readily available to typical research institutions.

Cases: Lattice QCD

Lattice Quantum Chromo Dynamics calculations are solving fundamental problems in particle and nuclear physics with large-scale computer calculations. The presentation includes performance evaluation and MPI profiling information.

For questions or comments, please contact info@hpcadvisorycouncil.com

@mellanox.com">HPC@mellanox.com