Application Performance on High-End and Commodity-class Computers

14/10/2001


Click here to start


Table of Contents

Application Performance on High-End and Commodity-class Computers

Outline

Scientific Computing - Cost Effectiveness & Dependencies

Cluster Computing -Where Do Clusters Fit?

What is Driving Commodity Clusters?

Daresbury Beowulf Project

High-end Commodity-based Systems

Joint Research Equipment Initiative (JREI)

Beowulf Chemistry Sites

Technical Progress in 2000/2001

Commodity Systems (CSx) Prototype / Evaluation Hardware

High-End Systems

SPEC CPU 2000 - Floating point Benchmark Suite (SPECfp2000)

SPEC CPU 2000 - SPECfp2000 Values relative to Compaq Alpha ES40/833

The GAMESS-UK Benchmark

STREAM: Measured Sustainable Memory Bandwidth in HPC (TRIAD)

Interconnect Comparison - MPI

Communications Benchmark PMB: Pallas MPI Benchmark Suite (V2.2)

PingPong Performance

MPI_allreduce Performance

Beowulf Prototype Systems: CS-0 and CS-1

Beowulf Prototype Systems: CS-2

QSNet Alpha/LINUX Cluster

Application Codes

DL_POLY: A Parallel Molecular Dynamics Simulation Package

DL_POLY Parallel Benchmarks (Cray T3E/1200)

DL_POLY: Cray/T3E, High-end and Commodity-based Systems I.

DL_POLY: Cray/T3E, High-end and Commodity-based Systems II.

DL_POLY: Cray/T3E, High-end and Commodity-based Systems III.

Vampir

CHARMM

Distributed Data SCF

GAMESS-UK

Global Arrays

PeIGS 3.0 Parallel Performance

Parallel Implementations of GAMESS-UK

GAMESS-UK ?SCF Performance Cray T3E/1200E, High-end and Commodity-based Systems

?SCF Performance - Cray T3E/1200E , SGI Origin3800/R14k-500 and Compaq AlphaServer SC/667

GAMESS-UK. DFT B3LYP Performance Cray T3E/1200, High-end and Commodity-based Systems

GAMESS-UK. DFT B3LYP Performance The Cray T3E/1200 and High-end Systems

Auxilliary Basis Coulomb Fit (I)

Auxilliary Basis Coulomb Fit (ii)

GAMESS-UK: DFT HCTH on Valinomycin. Impact of Coulomb Fitting: Cray T3E/1200, High-end and Commodity Systems

GAMESS-UK: DFT HCTH on Valinomycin. Impact of Coulomb Fitting: Cray T3E/1200, Cray Alpha Cluster/833 and Compaq AlphaServer SC/667

GAMESS-UK: DFT HCTH on Valinomycin. Speedups for both Explicit and Coulomb Fitting.

DFT HCTH Performance : Valinomycin

Case Studies - Zeolite Fragments

GAMESS-UK: DFT S-VWN Impact of Coulomb Fitting: Compaq AlphaServer SC /833

DFT Coulomb Fit - GAMESS-UK

DFT JFit Performance : Si26O37H36

DFT BLYP Gradient: Cray T3E/1200, IBM SP/WH2 and Beowulf Systems

MP2 Gradient Algorithms

Performance of MP2 Gradient Module Cray T3E/1200, High-end and Commodity- based Systems

SCF Analytic 2nd Derivatives Performance Cray T3E/1200, High-end and Commodity-based Systems

CRYSTAL98: Periodic SCF for MgO and TiO2

CRYSTAL - 2000

CASTEP - The UK Car-Parrinello Consortium

CASTEP 4.2 - Parallel Benchmark

CPMD - Car-Parrinello Molecular Dynamics

ANGUS: Combustion modelling (regular grid) The Cray T3E/1200, IBM SP/WH2 and Beowulf Systems

ANGUS: Combustion modelling (regular grid) Memory Bandwidth Effects: The IBM SP and Alpha Linux System

FLITE3D: An Industrial Aerospace Code

Atmospheric Climate Modelling

Beowulf Comparisons with the T3E & O3800/R14k-500

Collaborations: Performance Modelling

Performance Analysis of GA-based Applications using Vampir

Performance analysis of GA-based applications using the Vampir tool

Vampir Features

Vampir Displays

Timeline Display (Message Info)

Vampirtrace

Vampirtrace

Vampirtrace API

Instrumenting single-sided memory access

Runtime tracing options

GAMESS-UK / Si8O25H18 : 8 CPUs: One DFT Cycle

GAMESS-UK / Si8O25H18 : 8 CPUs

Performance Analysis of GA-based Applications using Vampir - Summary

Summary

Author: Martyn F. Guest

Email: m.f.guest@dl.ac.uk