root/trunk/README.lyx

#LyX 1.2 created this file. For more info see http://www.lyx.org/
\lyxformat 220
\textclass article
\language english
\inputencoding auto
\fontscheme default
\graphics default
\paperfontsize default
\papersize Default
\paperpackage a4
\use_geometry 0
\use_amsmath 0
\use_natbib 0
\use_numerical_citations 0
\paperorientation portrait
\secnumdepth 3
\tocdepth 3
\paragraph_separation indent
\defskip medskip
\quotes_language english
\quotes_times 2
\papercolumns 1
\papersides 1
\paperpagestyle default

\layout Section*

Beowulf Performance Suite (BPS)
\layout Standard

The Beowulf Performance Suite (BPS) was designed to provide a comprehensive
 and comparative way of measuring cluster performance.
 Although BPS contains many benchmarking programs, BPS is not designed to
 benchmark clusters.
 BPS is designed as an analysis tool to measure differences due to hardware
 or software changes onthe same cluster.
 In addition, successfully running all the tests provides some assurances
 that the cluster is configured properly.
 The suite provides a graphical user interface for running the programs
 as well as generating HTML output files.
 The following tests are available:
\layout Itemize

bonnie - hard drive performance
\layout Itemize

stream - memory performance
\layout Itemize

netperf - general network performance
\layout Itemize

netpipe - detailed network performance
\layout Itemize

nas - nas NASA parallel tests
\layout Itemize

unixbench - general Unix benchmarks
\layout Itemize

lmbench - micro Linux benchmarks
\layout Subsection*

Installing BPS
\layout Standard

If the BPS package, bps-1.0-8.i386.rpm (version may vary), is not installed
 on your cluster, the package can be found at ftp.plogic.com.
\layout Subsection*

Running in X windows
\layout Standard

The BPS provides an X windows interface.
 The interface can be started by executing the following command:
\layout LyX-Code

# xbps
\layout Standard

The desired set of benchmark(s) can then be selected from the list.
 The selected benchmarks can then be run by pressing the Run selected benches
 button.
 This will not work with Netpipe or Netperf as they require user settings.
 In addition, the NAS suite will only run on one node unless other settings
 are provided.
 After your benches have run, an HTML file of the results can be produced
 by clicking on the Create html file of results button.
 This will create a file called index.html in the logs folder which you can
 open with your browser.
\layout Subsubsection*

Settings
\layout Standard

Using the settings is the better way to run the benchmarks, and in the case
 of Netpipe and Netperf is necessary.
 The settings, however, are not much more complex than running the benches.
 
\layout Itemize

Log Directory - Sets the directory your bench logs will be placed into.
 Also, when creating an html file from the results, you must create the
 html file in the same directory as your log directory.
 Otherwise the program will not be able to find the files it needs to create
 the page.
\layout Itemize

Netperf and Netpipe - The settings for both of these look the same but are
 distinct, as Netpipe is different than Netperf.
 You must provide a valid receiver and sender node in both cases.
 The receiver node setting is the node that will be running the bench as
 a receiver, and the sender node is the one that will be running it as a
 sender.
 Since these benches run as benchmarks for the network communications between
 the two nodes, it is necessary to have both nodes specified.
 The interface is determined by the names of the hosts.
\layout Itemize

NAS - The default NAS settings are shown when you choose thisoption.
 The default settings will run the NAS suite on a single node for the S
 class size (i.e.
 workstation).
 However, if you have two or more nodes/processors, you can add them in
 the machine list.
 Because compilation is a part of the tests NAS runs, you can specify which
 compiler you want it to use.
 You should also select the MPI versionand number of processors you have.
 
\layout Itemize

Machine Info - This is standard information about your motherboard,memory,
 type of network interface, and the distribution of Linux youare using.
\layout Itemize

Prompting - This turns window prompts on/off.
\layout Subsubsection*

Finding out more about the benchmarks
\layout Standard

In the Help menu, you can find a one sentence synopsis of what each benchmark
 does by selecting Benchmark Info.
\layout Subsection*

Running in text mode
\layout Standard

To run bps in text mode enter:
\layout LyX-Code

# bps
\layout Standard

The following command line options are available:
\layout LyX-Code

# bps -h
\layout LyX-Code

Usage: /usr/bps/bin/bps <OPTIONS>
\layout LyX-Code

Options: 
\layout LyX-Code

-b bonnie++
\layout LyX-Code

-s stream
\layout LyX-Code

-f <send node>,<receive node> netperf to remote node
\layout LyX-Code

-p <send node>,<receive node> netpipe to remote node
\layout LyX-Code

-n <compiler>,<#processors), NAS parallel benchmarks
\layout LyX-Code

<test size>,<MPI>, compiler={gnu,pgi,intel}
\layout LyX-Code

<machine1,machine2,...> test size={A,B,C,dummy}
\layout LyX-Code

MPI={mpich,lam,mpipro}
\layout LyX-Code

-k keep NAS directory when finished
\layout LyX-Code

-u unixbench
\layout LyX-Code

-m lmbench
\layout LyX-Code

-l <log_dir> benchmark log directory
\layout LyX-Code

-w preserve existing log directory
\layout LyX-Code

-i <mboard manufacturer>, machine information
\layout LyX-Code

<mboard model>,<memory>,
\layout LyX-Code

<interconnect>,<linux ver>
\layout LyX-Code

-v show version
\layout LyX-Code

-h show this help
\layout Standard

If you are running the NAS tests from the bps script, you may tell it to
 keep the NAS directory (npb) in your log directory.
 This can be useful in tracking down compile errors (see the ~npb/make.log).
 HTML output files can be produced from the results files in <log directory>
 by executing:
\layout LyX-Code

# bps-html <log directory>
\layout Subsection*

Important Notes
\layout Itemize

All tests are archived in the src directory.
 
\layout Itemize

The bps suite is best run as a normal user, not root.
 Some of the tests (i.e.
 NAS parallel) will not run as root.
\layout Itemize

Not all features of the command line interface are possible with the GUI.
\layout Itemize

When using Netpipe and Netperf Benchmarks, rsh with no password must be
 permitted between the nodes.
\layout Itemize

Under normal operation, xbps will always preserve the existing log directory.
 This feature is to ensure previous results will not be overwritten.
 You can copy previous log files (from log directories) into the current
 log directory for bps-html conversion.
\layout Itemize

Also, the tests have been designed so that the BPS rpm only needs to be
 installed on the master node.
 For this to work, the BPS log directory must be mounted on all nodes (e.g.
 under /home).
 
\layout Itemize

When using the NAS Parallel Benchmarks it is advisable to use the the Paralogic
 MPI's which which have been tested to work with the NAS suite.
 Please see the NAS documentation for more information.
 Rather than limit potential BPS users, these are not made a part of the
 required packages list.
 The benchmark scripts have been written to rely on the three environment
 variables (for LAM-MPI, MPI-PRO, and MPICH).
 If you are having problems with the NAS benchmarks, extract the npb.tar.gz
 archive in the/usr/bps/src directory and try running the scripts manually.
 Consult the README.plogic file for more information.
 Also, if you wish to use the Portland Group or the Intel Compilers make
 sure they are properly configured.
\layout Subsection*

In Case of Problems
\layout Standard

The BPS suite is a collection of many tests.
 You should have minimal or no problems with the single machine tests.
 As more machines become involved with the tests, there is room for more
 configuration errors to arise.
 If a test does not run, check the test_name.log file in the log directory.
 In the case of the NAS tests, the resultsare in the form npb.COMPILER.MPI.CLASS.PRO
CESSORS.
 
\layout Standard

In general, if you have problems with a test it may be best to run it from
 the command line.
 In the case of the NAS suite, the -k option will keep the npb directory
 in the log directory so you can run the tests more directly by using the
 run_suite script in the npb directory.
 Also the README.plogic file in the npb directly should provide more information
 on how the tests are run and how to resolve possible problems.
\layout Subsection*

NASA Benchmark Suite (details)
\layout Standard

The NAS suite can be run from the command line.
 The following options are required.
 You also need to list the machine names in npb/cluster/machines file (one
 per line).
 This file is used by MPI to start your programs.
 In the npb directory, run
\layout LyX-Code

# run_suite -h
\layout LyX-Code

Usage: ./run_suite <OPTIONS>
\layout LyX-Code

Options:
\layout LyX-Code

run_suite,v 0.3 2/12/2002
\layout LyX-Code

Usage: ./run_suite <OPTIONS>
\layout LyX-Code

Options:
\layout LyX-Code

-v verbose output from make stage (default=make.log)
\layout LyX-Code

-c <compiler> compiler (gnu/pgi/intel)
\layout LyX-Code

-n <processors> number of processors
\layout LyX-Code

-t <test> test size (A,B,C,S)
\layout LyX-Code

-m <mpi> mpi version(lam,mpich,mpipro,dummy)
\layout LyX-Code

-o only build programs
\layout LyX-Code

-h show this help
\layout LyX-Code

To run on a single CPU use: '-c gnu -n 1 -t S -m dummy'
\layout Subsubsection*

The following is a description of the NAS tests.
\layout Description

BT is a simulated CFD application that uses an implicit algorithm to solve
 3dimensional (3D) compressible NavierStokes equations.
 The finite differences solution to the problem is based on an Alternating
 Direction Implicit (ADI) approximate factorization that decouples the x,
 y, and z dimensions.
 The resulting systems are BlockTridiagonal of 5x5 blocks and are solved
 sequentially along each dimension.
 
\layout Description

SP is a simulated CFD application that has a similar structure to BT.
 The finite differences solution to the problem is based on a BeamWarming
 approximate factorization that decouples the x, y, and z dimensions.
 The resulting system has scalar Pentadiagonal bands of linear equations
 that are solved sequentially along each dimension.
 
\layout Description

LU is a simulated CFD application that uses symmetric successive overrelaxation
 (SSOR) method to solve a seven block diagonal system resulting from finite
 difference discretization of the NavierStokes equations in 3D by splitting
 to into block Lower and Upper triangular systems.
 
\layout Description

FT contains the computational kernel of a 3D fast Fourier Transform (FFT)based
 spectral method.
 FT performs three one dimensional (1D) FFT's, one for each dimension.
 
\layout Description

MG uses a Vcycle MultiGrid method to compute the solution of the 3D scalar
 Poisson equation.
 The algorithm works continuously on a set of grids that are made between
 coarse and fine.
 It tests both short and long distance data movement.
 
\layout Description

CG uses a Conjugate Gradient method to compute an approximation to the smallest
 eigenvalue of a large, sparse, unstructured matrix.
 This kernel tests unstructured grid computations and communications by
 using a matrix with randomly generated locations of entries.
 
\layout Description

EP is an Embarrassingly Parallel benchmark.
 It generates pairs of Gaussian random deviates according to a specific
 scheme.
 The goal is to establish the reference point for peak performance of a
 given platform.
 
\layout Subsection*

Additional benchmark information:
\layout Itemize

General Information http://www.plogic.com/bps
\layout Itemize

bonnie++ Hard drive performance - http://www.coker.com.au/bonnie++
\layout Itemize

stream Memory performance http://www.cs.virginia.edu/stream
\layout Itemize

netperf General network performance http://www.netperf.org/netperf/NetperfPage.html
\layout Itemize

netpipe Detailed network performance 
\layout Itemize

http://www.scl.ameslab.gov/Projects/ClusterCookbook/nprun.html
\layout Itemize

unixbench General Unix benchmarks http://www.linuxdoc.org/HOWTO/Benchmarking-HOWTO.
html
\layout Itemize

LMbench Low level benchmarks 
\layout Itemize

http://www.bitmover.com/lmbench
\layout Itemize

NAS Parallel tests http://www.nas.nasa.gov/Software/NPB
\the_end