
CS 594 Scientific Computing for
Engineers: Spring 2010
– 3 Credits This class is part of the Interdisciplinary Graduate Minor in Computational
Science. See IGMCS for details.
Wednesdays from 1:30 – 4:15 Prof.
Jack Dongarra with help from Profs. George Bosilca,
Heike Jagode, Jakub Kurzak,
and Stan Tomov
Email: dongarra@eecs.utk.edu Phone: 8659748295 Fax: 8659748296 Office hours: Wednesday 11:00  1:00, or by
appointment TA: Donald Newell
<donaldnewell08@gmail.com> TA’s Office : TA’s
Office Hours: Wednesday’s 10:00 – 12:00 or by appointment There will be four major aspects of the course: · Part I will start with current trends in highend computing systems and environments, and continue with a practical short description on parallel programming with MPI, OpenMP, and pthreads. · Part II will illustrate the modeling of problems from physics and engineering in terms of partial differential equations (PDEs), and their numerical discretization using finite difference, finite element, and spectral approximation. · Part III will be on solvers: both iterative for the solution of sparse problems of part II, and direct for dense matrix problems. Algorithmic and practical implementation aspects will be covered. · Finally in Part IV, various software tools will be surveyed and used. This will include PETSc, Sca/LAPACK, MATLAB, and some tools and techniques for scientific debugging and performance analysis. The grade would be based on homework, a midterm project, a final project, and a final project presentation. Topics for the final project would be flexible according to the student's major area of research. Class Roster If
your name is not on the list or some information is incorrect, please send
mail to the TA: And the course mailing list: cs594parallelstudents@cs.utk.edu Book for the Class: The Sourcebook of Parallel Computing, Edited
by Jack Dongarra, Ian Foster, Geoffrey Fox, William Gropp, Ken Kennedy, Linda
Torczon, Andy White, October 2002, 760 pages, ISBN 1558608710, Morgan
Kaufmann Publishers. Lecture Notes: (Tentative outline of the
class)
Introduction
to High Performance Computing Read Chapter
1, 2, and 9 Homework 1 (due January 27,
2010)
Projection and its importance in scientific
computing Homework 2
(due February 2, 2010)
Parallel programming paradigms and their
performances Homework 3 (due February 11, 2010)
Message
Passing (MPI1 the basics) Parallel programming paradigms and their
performances 2 Read Chapter
11
Message Passing Interface
(MPI) MPI Dynamic Processes
MPI MPI
I/O
Read Chapter 9
HPC Performance Issues
and Systems Read Chapter 3
Read
Chapter 15
Read Chapter 3 Toward an Optimal Algorithm for Matrix Multiply Read Chapter 20 March 10 – Spring Break
Discretization of PDEs and tools for the parallel
solution of the resulting systems
and Mesh Generation and Load Balancing 10.
March 24 (Dr. Tomov) Sparse matrices and optimized parallel
implementations
Iterative Methods in Linear Algebra Read
Chapter 20 and 21 12. April
7 (Dr. Tomov) Iterative Methods in Linear Algebra (Continued)
Discussion
on CUDA and the Fermi Architecture
[see http://www.nvidia.com/object/cuda_home_new.html
and http://www.nvidia.com/object/fermi_architecture.html ]
Additional material Matrix3x3.cpp Matrix.cpp kernels.cu hw8.cpp
Discussion on Linear Algebra for New Architectures Discussion on Final Projects
Some disasters attributable to bad numerical computing Read Chapter 20, Bailey’s
paper on “12 ways to fool …”
Read
Chapter 21
Class
Final reports Order
of presentation: Here
are some ideas for projects: o Projects
and additional
projects. Additional Reading Materials
Message
Passing Systems.
The PVM
home page. Other
useful reference material
·
Here’s a pointer to specs on various processors: http://www.cpuworld.com/CPUs/index.html http://www.cpuworld.com/sspec/index.html http://processorfinder.intel.com A good introduction to message passing
systems.
``Message Passing Interfaces'', Special issue of Parallel Computing
, vol 20(4), April 1994. A paper by members of the PVM team on the differences between PVM
and MPI. Geist, G.A, J.A. Kohl, P.M. Papadopoulos, `` PVM and MPI: A
Comparison of Features '', Calculateurs Paralleles , 8(2), pp.
137150, June, 1996. Papers by members of the MPI team on the differences between PVM
and MPI. ``Why are PVM and MPI So Different'', William Gropp and Ewing Lusk
(submitted to The Fourth European PVM  MPI Users' Group Meeting) and ``PVM and MPI are completely different'', William Gropp and Ewing
Lusk, to appear in the journal Future Generation Computer Systems, 1998. Ian Foster, Designing and Building Parallel Programs, see http://wwwunix.mcs.anl.gov/dbpp/ Alice
Koniges, ed., Industrial Strength Parallel Computing,
ISBN1558605401, Morgan Kaufmann Publishers, San Francisco, 2000. Ananth Gramma et al., Introduction
to Parallel Computing Michael Quinn, Parallel Programming, see http://web.engr.oregonstate.edu/~quinn/Comparison.htm David E. Culler & Jaswinder Pal Singh, Parallel
Computer Architecture, see http://www.cs.berkeley.edu/%7Eculler/book.alpha/index.html George Almasi and Allan Gottlieb, Highly Parallel
Computing Standard Books on Message Passing
``MPI  The Complete
Reference, Volume 1, The MPI1 Core, Second Edition'', ``Using MPI,'' ``MPI: The Complete Reference  2nd Edition:
Volume 2  The MPI2 Extensions'', Online Documentation and Information about
Machines ·
Overview of Recent
Supercomputers, Aad J. van der Steen and Jack J. Dongarra, 2007.
·
Catalog
of Commercial Hardware and Software Vendors Other Parallel
Information Sites ·
NHSE
 National HPCC Software Exchange ·
Netlib
Repository at UTK/ORNL ·
LAPACK
·
GAMS
 Guide to Available Math Software ·
Supercomputing
& Parallel Computing: Conferences ·
Supercomputing
& Parallel Computing: Journals ·
High
Performance Fortran (HPF) reports ·
High
Performance Fortran Resource List ·
Major Science Research
Institutions from Caltech ·
Message
Passing Interface (MPI) Forum ·
High
Performance Fortran Forum ·
OpenMP
·
PVM
·
DoD
High Performance Computing Modernization Program ·
DoE
Accelerated Strategic Computing Initiative (ASC) ·
National
Computational Science Alliance Related Online
Textbooks · Templates
for the Solution of Linear Systems: Building Blocks for Iterative Methods,
SIAM Publication, Philadelphia, 1994. · PVM
 A Users' Guide and Tutorial for Networked Parallel Computing, MIT Press,
Boston, 1994. · MPI
: A MessagePassing Interface Standard · LAPACK
Users' Guide (Second Edition), SIAM Publications, Philadelphia, 1995. · MPI:
The Complete Reference, MIT Press, Boston,
1996. ·
Using
MPI: Portable Parallel Programming with the MessagePassing Interface
by W. Gropp, E. Lusk, and A. Skjellum · Parallel
Computing Works, by G. Fox, R. Williams, and P. Messina
(Morgan Kaufmann Publishers) · Computational
Science Education Project · Designing and Building Parallel
Programs. A deadtree version of this book is
available by AddisonWesley. · High
Performance Fortran (HPF), a course offered by Manchester
and North High Performance Computing Training & Education Centre, United
Kingdom For performance analysis: · Raj Jain, The Art of Computer Systems
Performance Analysis. John Wiley, 1991. Papers on performance analysis tools: · Ruth A. Aydt, "The Pablo
SelfDefining Data Format," November 1997, click
here. · Jeffrey K. Hollingsworth, Barton P.
Miller, Marcelo J. R. Gongalves, Oscar Naim, Zhichen Xu and Ling Zheng,
"MDL: A Language and Compiler for Dynamic Program Instrumentation",
International Conference on Parallel Architectures and Compilation
Techniques, San Francisco, CA, November 1997, click
here. · Barton P. Miller, Mark D. Callaghan,
Jonathan M. Cargille, Jeffrey K. Hollingsworth, R. Bruce Irvin, Karen L.
Karavanic, Krishna Kunchithapadam and Tia Newhall. "The Paradyn Parallel
Performance Measurement Tools", IEEE Computer 28(11), (November 1995). click
here. · Steven T. Hackstadt and Allen D.
Malony, "Distributed Array Query and Visualization for High Performance
Fortran”, February 1996.
click
here. · Jerry Yan and Sekhar Sarukkai and
Pankaj Mehra, "Performance Measurement, Visualization and Modeling of
Parallel and Distributed Programs using the AIMS toolkit", Software
Practice and Experience 25(4), April 1995, 429461. click
here. Other Online Software and Documentation· Matlab documentation is available from several sources, most notably by typing ``help'' into the Matlab command window. A primer (for version 4.0/4.1 of Matlab, not too different from the current version) is available in either postscript or pdf. · Netlib, a repository of numerical software and related documentation · Netlib Search Facility, a way to search for the software on Netlib that you need · GAMS  Guide to Available Math Software, another search facility to find numerical software · Linear Algebra Software Libraries and Collections · LAPACK, stateoftheart software for dense numerical linear algebra on workstations and sharedmemory parallel computers. Written in Fortran. · CLAPACK,
a C version of LAPACK. · ScaLAPACK, a partial version of LAPACK for distributedmemory parallel computers. · LINPACK and EISPACK are precursors of LAPACK, dealing with linear systems and eigenvalue problems, respectively. · SuperLU is a fast implementations of sparse Gaussian elimination for sequential and parallel computers, respectively. · Sources of test matrices for sparse matrix algorithms · University of Florida Sparse Matrix Collection · Templates for the solution of linear systems, a collection of iterative methods, with advice on which ones to use. The web site includes online versions of the book (in html and postscript) as well as software. · Templates for the Solution of Algebraic Eigenvalue Problems is a survey of algorithms and software for solving eigenvalue problems. The web site points to an html version of the book, as well as software. · Updated survey of sparse direct linear equation solvers, by Xiaoye Li · MGNet is a repository for information and software for Multigrid and Domain Decomposition methods, which are widely used methods for solving linear systems arising from PDEs. · Resources for Parallel and High Performance Computing · Millennium a UC Berkeley campuswide parallel computing resource · Resources for CS 267, Applications of Parallel Computers · ACTS (Advanced CompuTational Software) is a set of software tools that make it easier for programmers to write high performance scientific applications for parallel computers. · PETSc: Portable, Extensible, Toolkit for Scientific Computation · NHSE  National High Performance Computing and Communications Software Exchange, pointers to related work across the country. · Issues related to Computer Arithmetic and Error Analysis · Efficient software for very high precision floating point arithmetic · Notes on IEEE Floating Point Arithmetic, by Prof. W. Kahan · Other notes on arithmetic, error analysis, etc. by Prof. W. Kahan · Report on arithmetic error that cause the Ariane 5 Rocket Crash · The IEEE floating point standard is currently being updated. To find out what issues the standard committee is considering, look here. 4/29/2010


