HELP! * GREY=local Full HTML for

LOCAL foilset Conclusions of Working Group 2 at Second Pasadena Workshop on System Software and Tools for HPCC

Given by Geoffrey C. Fox at Second Pasadena Workshop on HPCC Software on January 10-11,1995. Foils prepared January 21,1995
Abstract * Foil Index for this file

See also color IMAGE
We discussed 12 application areas -- "vignettes" -- of which 3 -- all from industry -- are contained in final short report
  • Real Time Embedded Systems
  • Structural Dynamics
  • Manufacturing and Design
We discussed at length the impact of "non-technical" issues such as setting up:
  • End User -- ISV -- MPP Vendor Business Model with
  • Appropriate Academia -- Government Synergy
These non-technical issues lead to technical points which ensure a better more predictable HPCC software development environment

Table of Contents for full HTML of Conclusions of Working Group 2 at Second Pasadena Workshop on System Software and Tools for HPCC

1 Second Pasadena Workshop: Working Group 2 - Characteristics of Scientific and Engineering Applications.
2 Overview of WG2 Deliberations
3 Charter: Question 2.1: What are the characteristics of scientific and engineering applications and algorithms that require high-performance computer systems?
4 Charter: Question 2.2: What are the highest priority issues for system software and tools implied by these applications and algorithms?
5 Charter: Question 2.3: How will these applications and algorithms contribute to the long term commercial viability of high-performance computing systems?
6 Initial Summary of Technical Points
7 Enterprise Models:
(Forms of Industry, Government and Academic (IGA) collaboration)
8 Comments on Enterprise Models:
9 Another Non Technical Issue -- What is the Community ?
10 Initial Vignettes Identified
11 Action Items:
1) Support viable base enterprise model with platform independent standards scaling from workstations to MPPs with tools using these standards
12 Action Items:
2) Change Government funding as implied by the viable enterprise models?
13 Action Items:
3) Broaden application base outside traditional science and technology applications

This table of Contents Abstract

HELP! * GREY=local HTML version of LOCAL Foils prepared January 21,1995

Foil 1 Second Pasadena Workshop: Working Group 2 - Characteristics of Scientific and Engineering Applications.

From Working Group 2 at Pasadena2 Workshop Second Pasadena Workshop on HPCC Software -- January 10-11,1995. * See also color IMAGE
Full HTML Index
Chair: Geoffrey Fox
Co-Chair: Andy White
Secretary: Ken Hawick
January 10-12,1995 Pasadena
HELP! * GREY=local HTML version of LOCAL Foils prepared January 21,1995

Foil 2 Overview of WG2 Deliberations

From Working Group 2 at Pasadena2 Workshop Second Pasadena Workshop on HPCC Software -- January 10-11,1995. * See also color IMAGE
Full HTML Index
We discussed 12 application areas -- "vignettes" -- of which 3 -- all from industry -- are contained in final short report
  • Real Time Embedded Systems
  • Structural Dynamics
  • Manufacturing and Design
We discussed at length the impact of "non-technical" issues such as setting up:
  • End User -- ISV -- MPP Vendor Business Model with
  • Appropriate Academia -- Government Synergy
These non-technical issues lead to technical points which ensure a better more predictable HPCC software development environment
HELP! * GREY=local HTML version of LOCAL Foils prepared January 21,1995

Foil 3 Charter: Question 2.1: What are the characteristics of scientific and engineering applications and algorithms that require high-performance computer systems?

From Working Group 2 at Pasadena2 Workshop Second Pasadena Workshop on HPCC Software -- January 10-11,1995. * See also color IMAGE
Full HTML Index
Non-technical Characteristics of broad significance such as academia vs industry, and/or
Technical characteristics of specific applications (regular vs irregular)
HELP! * GREY=local HTML version of LOCAL Foils prepared January 21,1995

Foil 4 Charter: Question 2.2: What are the highest priority issues for system software and tools implied by these applications and algorithms?

From Working Group 2 at Pasadena2 Workshop Second Pasadena Workshop on HPCC Software -- January 10-11,1995. * See also color IMAGE
Full HTML Index
Identified several specific technical issues
HELP! * GREY=local HTML version of LOCAL Foils prepared January 21,1995

Foil 5 Charter: Question 2.3: How will these applications and algorithms contribute to the long term commercial viability of high-performance computing systems?

From Working Group 2 at Pasadena2 Workshop Second Pasadena Workshop on HPCC Software -- January 10-11,1995. * See also color IMAGE
Full HTML Index
Coupled to "broad non-technical issues" identified in Question 2.1.
HELP! * GREY=local HTML version of LOCAL Foils prepared January 21,1995

Foil 6 Initial Summary of Technical Points

From Working Group 2 at Pasadena2 Workshop Second Pasadena Workshop on HPCC Software -- January 10-11,1995. * See also color IMAGE
Full HTML Index
1) need for better debuggers, profilers, performance monitoring tools
2) need for more stable operating systems
3) need for tools to aid in code migration to parallel systems, whether it be in the form of libraries, or other software engineering tools.
4) need to reduce the latencies due to system software
5) need for looking at exciting and innovative applications areas, (to help the HPCC industry by stimulating new demands). This might involve very data intensive applications (in contradistinction to compute intensive ones) but also harder and more complex problems, irregular data structures and less obviously load balanceable problems.
HELP! * GREY=local HTML version of LOCAL Foils prepared January 21,1995

Foil 7 Enterprise Models:
(Forms of Industry, Government and Academic (IGA) collaboration)

From Working Group 2 at Pasadena2 Workshop Second Pasadena Workshop on HPCC Software -- January 10-11,1995. * See also color IMAGE
Full HTML Index
1) Viable base model: Build HPCC software on an internally viable base such as distributed computing or the WWW.
2) Internally consistent model: areas where business case for HPCC is internally viable (eg decision support)
3) Partnership model: Government supported teams collaborating with industry teams (eg oil and gas)
4) Pulse/Seed support model: IGA teams to develop applications (eg Europort, IBM, TMC,...)
5) Ongoing support model: of areas of national importance, but without identified commercial markets (NSA, Weapons, QCD)
6) Dual benefit model: Government market bootstraps viable commercial market or vice versa.
HELP! * GREY=local HTML version of LOCAL Foils prepared January 21,1995

Foil 8 Comments on Enterprise Models:

From Working Group 2 at Pasadena2 Workshop Second Pasadena Workshop on HPCC Software -- January 10-11,1995. * See also color IMAGE
Full HTML Index
Categories are not rigid but (six) approximately defined regions in complex multidimensional space (they could be merged or overlapped). Different application areas have different investment strategies
Different applications have a different mix of metrics such as:
  • Economic value;
  • Contribution to fundamental knowledge;
  • National Security;
  • Quality of life.
  • (as examples)
HELP! * GREY=local HTML version of LOCAL Foils prepared January 21,1995

Foil 9 Another Non Technical Issue -- What is the Community ?

From Working Group 2 at Pasadena2 Workshop Second Pasadena Workshop on HPCC Software -- January 10-11,1995. * See also color IMAGE
Full HTML Index
Need to involve a larger group of non HPCC communities
For instance, most of the messages on networks are
  • MIME (Email/World Wide Web) or eventually ATM (of one or another adaptation layer AAL)
  • Not PVM/MPI
But MPI standrards set internally to HPCC and did not explicitly involve ATM/Internet community/standard processes
HPF focusses on regular multidimensional arrays in an excellent standards forum that ignores
  • VRML which is a fascinating 3D (irregular) datastructure which surely needs HPCC and will greatly broaden relevance of HPCC as used in interactive simulations of virtual worlds
Need HPVRML and a broader community
HELP! * GREY=local HTML version of LOCAL Foils prepared January 21,1995

Foil 10 Initial Vignettes Identified

From Working Group 2 at Pasadena2 Workshop Second Pasadena Workshop on HPCC Software -- January 10-11,1995. * See also color IMAGE
Full HTML Index
NASTRAN
Real-time embedded systems
Aerospace manufacturing
Crisis management
Nuclear Weapons
Environmental Modeling
Mission to Planet Earth
Data Intensive Applications
High end CFD
Centric
Computational Chemistry
QCD
HELP! * GREY=local HTML version of LOCAL Foils prepared January 21,1995

Foil 11 Action Items:
1) Support viable base enterprise model with platform independent standards scaling from workstations to MPPs with tools using these standards

From Working Group 2 at Pasadena2 Workshop Second Pasadena Workshop on HPCC Software -- January 10-11,1995. * See also color IMAGE
Full HTML Index
Currently the tail is wagging the dog - the BIG dog?
What is the market area that is big enough upon which to base viable HPCC standards (eg SMP, distributed systems or WWW)?
What are the top three standards?
HELP! * GREY=local HTML version of LOCAL Foils prepared January 21,1995

Foil 12 Action Items:
2) Change Government funding as implied by the viable enterprise models?

From Working Group 2 at Pasadena2 Workshop Second Pasadena Workshop on HPCC Software -- January 10-11,1995. * See also color IMAGE
Full HTML Index
Industry/market application driven
Tools and tools funding which are implied by viable enterprise models
Viable base and Pulse/Seed funding enterprise models appear to be high priority
HELP! * GREY=local HTML version of LOCAL Foils prepared January 21,1995

Foil 13 Action Items:
3) Broaden application base outside traditional science and technology applications

From Working Group 2 at Pasadena2 Workshop Second Pasadena Workshop on HPCC Software -- January 10-11,1995. * See also color IMAGE
Full HTML Index
Low latency (eg real-time)
design and manufacturing
data intensive (eg crisis management)
business
event-driven modeling and simulation
Northeast Parallel Architectures Center, Syracuse University, npac@npac.syr.edu

If you have any comments about this server, send e-mail to webmaster@npac.syr.edu.

Page produced by wwwfoil on Mon Feb 17 1997