MCS 572 Introduction to Supercomputing
----- SPRING 2003 -----

Quick Links:

    |   PSC TCS   |   NCSA Platinum   |   UIC Argo (TOC)   |   Cluster Computing Info.   |  
    |   Platinum/Argo Project   |   TCS Project   |   TCS Starter Problem   |   Theoretical Homework   |  
    |   MPI Info.   |   MPI Examples   |  
    |   NCSA Platinum Cluster Local Guide   |   Argo Cluster Local Guide   |   PSC TSC Cluster Local Guide   |  
    |   Cray T90/T3E Guide   |  
    |   Web-Texts   |   Super Tools   |   Unix   |   Fortran   |   C/C++   |   SciVis   |  

• E-Mail: hanson A T uic edu

• Web Page: http://www.math.uic.edu/~hanson/

• Office Hours: 1:35-2:50pm MW 718 SEO (by arrangement; email preferable);

Catalog description: Introduction to supercomputing on vector, parallel and massively parallel processors; architectural comparisons, parallel algorithms, vectorization techniques, parallelization techniques, actual implementation on real machines (Crays, Convex, and others).

Prerequisites: Some numerical experience such as MCS 471 Numerical Analysis or MCS 571 Numerical Methods for Partial Differential Equations or consent of the instructor. Graduate standing.

Semester Credit hours: 4

List of Topics --- Hours

• Introduction to advanced scientific computing. --- 3 hours.
• Comparison of serial, parallel and vector architectures. --- 3 hours.
• Performance measures and models of performance. --- 3 hours.
• Pure parallel algorithms and data dependencies. --- 3 hours.
• Optimal code design. --- 3 hours.
• High Performance Programming Languages (eg, F90, HP/C). --- 2 hours.
• Loop optimization by reformulation. --- 6 hours.
• Code implementation on parallel computing clusters. --- 13 hours.
• Parallel Programming Interfaces (eg, MPI, PVM) --- 6 hours.
• Block decomposition and interation methods. --- 3 hours.
• Total. --- 45 hours.

Grade Basis: Homework, Starter Problems, Computer Project Reports (2-3) and Project Presentations.

Recommended References (lecture based course with no required text):

• F. B. Hanson, Local Supercomputing Training in the Computational Sciences Using Remote National Centers, Future Generation Computer Systems: Special Issue on Education in the Computational Sciences, accepted, 25 pages in ms., August 2002.

• F. B. Hanson, MCS 572 Local User's Guide to UIC ACCC Argo Cluster for Spring 2003.

• F. B. Hanson, MCS 572 Local User's Guide to NCSA Platinum Cluster for Spring 2003.

• F. B. Hanson, MCS 572 Local User's Guide to PSC TSC Cluster for Spring 2003.

• F. B. Hanson, A Real Introduction to Supercomputing, in Proc. Supercomputing '90, pp. 376-385, Nov. 1990

• F. B. Hanson, MCS572 UIC Cray User's Local Guide to NPACI-SDSC Cray T90 Vector Multiprocessor and T3E Massively Parallel Processor, for Fall 2000.

• A. Barry Wilkinson and C. Michael Allen, Parallel Programming: Techniques and Application Using Networked Workstations and Parallel Computers, Prentice Hall PTR, Upper Saddle River, NJ, 1999.
  1. See Amazon.com for new and used paperback editions from $47
  2. See Wilkinson's Lecture Notes.

• Rajkumar Buyya, Monash University Australia, Cluster Computing Information
  1. Rajkumar Buyya (editor)and Hai Jin (slide author), High Performance Cluser Computing: Architures and Systems -- Lecture Notes for selected chapters ( MS PowerPoint Viewer needed, but available free on-line).
  2. Rajkumar Buyya, High Performance Cluser Computing Tutorials, used at International Conferences ( MS PowerPoint Viewer needed, but available free on-line for MS Windows).

• Kai Hwang, University of Southern California, Basic Network-Based Cluster Computing, lecture notes.

• Jack J. Dongarra, Ian S. Duff, Danny C. Sorensen and Henk A. van der Vorst, Solving Linear Systems on Vector and Shared Memory Computers, SIAM, 1990. (A new edition is supposed to come out during November 1998 according to Dongarra and SIAM, entitled Numerical Linear Algebra for High-Performance Computers, $37 list price ($26.60 member), order code SE07).

• Local User's Guide to PSC TSC Cluster for MCS 572 Spring 2003.

• Local User's Guide to NCSA Platinum Cluster for MCS 572 Spring 2003.

• PSC Terascale Computing System (TCS) Prototype, Pittsburgh Supercomputing Center, 256 Processors (64 Compaq Alphaserver ES40 nodes with 4 processors per node), Class Account Grant SEE030003P Approved 20 Jan 2003.

  • Terascale Computing System Summary, Schematic Slide used 22 Jan 2003.

  • TCS Compaq AlphaServer SC System Overiew, HP Workshop notes (you may need an AFS password for these nice PowerPoint Slides and will need a MS PowerPoint Viewer for MS Windows)

  • Parallel Programming Techniques, Online PSC Workshop Lecture Notes in Powerpoint ( Free MS PowerPoint Viewer):
    • Introduction to the PSC and Parallel Computing
    • Parallel Computing Overview
    • Compiling and Job Submission
    • MPI Basics (For HTML version, see MPI Message Passing Basics).
    • Message Passing: Code Development (For HTML version, see MPI Laplace Code Development).
    • Performance Optimization
    • TotalView Debugger

• NCSA Platinum IA32 Linux Cluster, National Center for Supercomputing Applications, UIU, Urbana, 968 processors (484 dual processor compute nodes of Intel Pentium III 1GHz chips), Class Account Grant ASC030009N (nfa) Approved 17 Jan 2003.

• Local User's Guide to UIC ACCC Argo Cluster for MCS 572 Spring 2003.

• ACCC Argo Beowulf Cluster, UIC ACCC 16 Compute Node (AMD 1600 XP 1.4GHz Athlon) cluster.
  
  
  1. ARGO: General InformationACCC User Basic Information Notes
     • Argo Beowulf Hardware Overview
       
     • Argo Pictures
       
     • ACCC Beowulf Hardware Details
  2. ARGO: Available Software
     
  3. ARGO: Getting Started and Running Jobs
     
     
     • Portable Batch System (PBS)
       
     • Message Passing Interface System (MPI) {use mpimon, not mpirun}
       
     • (1) Set Up a ScaMPI Environment {set paths if csh}
       
     • (2) Compile and Link {compiler options}
       
     • (3) Running a Job in Batch (scasub)
  4. ARGO: More on Running Parallel MPI Jobs
     
     
     • Argo Beowulf Cluster: MPI Commands
  5. ARGO: Glossary
     
  6. ARGO: Other References
     
  7. ACCC User Basic Information Notes for Argo
     
  8. ACCC User X-Windows GUI Passthrough Notes for Argo

• Possibly an alternate cluster.

• Trends in Cray Supercomputer versus Killer Micros Performance, from Buyya (Chapt. 1, p. 84), handed out 15 January 2003.

• Clusters with Myrinet Networks and More, HPCwire News, emailed to Class 17 January 2003.

• Top500 NCSA Platinum Linux Cluster Summary, 22 Jan 2003.

• Top500 Cluster Short List with top 10 plus some Academic Clusters, 22 Jan 2003.

• Kai Hwang's "Basic Network-Based Cluster Computing," Full Set of Slides, But Only a Selection of Lecture Notes Used, 22 Jan 2003.

• Raj Buyya's "Cluster Computing Architectrue," Single Slide used, 22 Jan 2003.

• Alphaserver SC Programmer's Quick Reference Guide, copy handed out 12 Feb 2003? (Note: Compact list of F90 and C command and parallel optimization options for the PSC TCS Cluster; print in Landscape mode.)

• MPI Reference Card (13 Compact Pages for MPICH (C)).

• .TCS Compaq AlphaServer SC System Overiew, HP Workshop notes (you may need an AFS password for these nice PowerPoint Slides and will need a MS PowerPoint Viewer for MS Windows) presented 17 Feb 2003.

On-Line Web Supercomputing Texts

Supplementary Supercomputing Texts

Additional Supercomputing Literature

• Top 500 Supercomputer Full List

• Gunter's Full List of Most Powerful Supercomputer SITES

• MCS572 Brief Top 500 Supercomputer List (Top 10, SDSC and NCSA only)

• MCS572 Brief List of Most Powerful Supercomputer SITES (Top 10, SDSC and NCSA only)

• MCS572 Log-Log Graph of #Processors versus #MegaFlops/Processor

Supercomputing Tools HTML Page

(MathLab, Maple, Octave, UNIX, Fortran, C, and Numerics)

Numeric and Symbolic Tools HTML Help Page

NPACI-SDSC and Cray Information:

• MCS572 UIC Cray User's Local Guide to NPACI-SDSC Cray T90 Vector Multiprocessor and T3E Massively Parallel Processor
  Cray T90 & T3E Local Guide

• NPACI-SDSC and Cray Information
  (Introductory Selections for MCS572

• NPACI High Performance Computing Asynchronous Distance Training
  NPACI HPC Distance Training Homepage
  {Caution: RealG2 player is required for viewing, and more.}

NCSA Information:

• NCSA SGI Cray Origin2000 Information:
  NCSA SGI Cray Origin2000 Homepage

General MPI Information:

• Introductory MPI Selections for MCS572

• Introductory MPI Examples for MCS572

General PVM Information:

• Introductory PVM for MCS572

HP9000/800 V2250 (new borg) Information:

UIC/MCS572 "borg" Local Users Guide in HTML

HP Convex SP1200 (old borg)Information:

• MCS572 UIC HP-Convex SPP1200/XA-16 Scalable Parallel Processor User's Local Guide
  
  UIC/MCS572 "old borg" Exemplar Local User's Guide (Fall97 MCS572 version ready)
• UIC HP-Convex Exemplar SPP1200/XA-16
  
  UIC Computer Center "borg" Exemplar Guide
• HP-Convex Exemplar SPP1600 (replaces SPP1200) Technical Summary
  
  Exemplar Technical Summary
• HP COnvex SPP1200 to SPP1600 Comparison
  (since SPP1200 Technical Summary is no longer availble).
  
  SPP1200 to SPP1600 Comparison
• HP-Convex Exemplar C and Fortran 77 Programmer's Guide
  
  Exemplar Programmer's Guide
• NCSA HP-Convex Exemplar Homepage.
  
  • NCSA HP-Convex Exemplar SPP1200/XA-64 Homepage. (NCSA SPP1200 has been replaced by the SPP2000 with faster CPUs)
  • NCSA HP-Convex Exemplar SPP-2000-64 (X-Class) Homepage.

PSC and Cray Information:

• MCS572 Introduction:
  Introductory Selections for MCS572

• PSC T3D to T3E Upgrade Migration Information:
  Cray T3E News