NPACI Partnership

1
NPACI Partnership

• Paul Messina
• Chief Architect, NPACI
• messina_at_cacr.caltech.edu
• Peter Arzberger
• NPACI Executive Director
• External Visiting Committee
• January 29, 1999

2
Partnership goal

• Creating a ubiquitous, continuous, pervasive HPCC
  infrastructure
• Infrastructure people, software, and hardware

3
Design of the partnership

• Based on a vision of new capabilities for CSE
• Selected, for infrastructure development,
  technology areas that are -- or promise to be --
  crucial for many applications
• Identified areas with important applications and
  in which NPACI partners have particular strengths
  to guide development

4
Partnership Organizing PrincipleMultidisciplina
ry Thrusts

• Application Thrusts
• Molecular Science
• Neuroscience
• Earth Systems Science
• Engineering

• Technology Thrusts
• Metasystems
• Programming Tools Environments
• Data-intensive Computing
• Interactive Environments

5
NPACI management team
Executive Committee

Leadership Team

• Click to edit Master text styles
• Second level
• Third level

Sid Karin Director
Peter Arzberger Executive Director
Paul Messina Chief Architect
Susan Graham Chief Computer Scientist
Peter Taylor Chief Applications Scientist
Wayne Pfeiffer Deputy Director
Greg Moses EOT Thrust Leader
Applications Thrust Leaders Russ Altman,
MoleSci Mark Ellisman, Neuro, Cross-Dis
Coord Bernard Minster, ESS Tinsley Oden, Eng
Enabling Technologies Thrust Leaders Andrew
Grimshaw, Meta Joel Saltz, PTE Reagan Moore,
DICE Arthur Olson, IE
Member at Large Bill Martin Jim Pool
6
Multidisciplinary approach

• Thrusts are our organizing principle
• Provide expertise and focus for individual areas
• Enhanced infrastructure results from applications
  push and technology pull
• Projects provide multidisciplinary setting for
  identifying, implementing, and testing new
  infrastructure
• Projects are composed of applications and
  technology people
• Multidisciplinary activities are a contact sport!

7
Projects implement the multidisciplinary approach

• Application thrusts identify
• Aspects of the HPC technologies that limit
  progress in computational science engineering
  work
• Specific problems not solvable with current
  infrastructure that would be enabled with
  enhanced infrastructure

8
Projects meld applications and technology
Brain databases
Data-IntensiveComputingNeuroscience
MetasystemsEarth SystemsScience
Global climate modeling
9
Project mode of operation

• Select specific CSE application problem to be
  solved
• Proves new infrastructure really works
• Exciting science and engineering results motivate
  others to use the new infrastructure
• Formulate plans with timelines and milestones
  for developing the targeted capability

10
Resource and EOT thrusts

• Resources
• Provide new capabilities for pursuing CSE and
  Computer Science
• Enable production use of new technologies
  (hardware and software) for applications
• Education, Outreach, and Training
• Enhance and enlarge the people component of
  infrastructure
• Dovetail with technology and applications thrusts
• Explore the use of HPCC in new communities

11
Enhancing the infrastructure

• Technology thrusts review solutions developed
  across applications
• Generalize what we learn and develop for the
  benefit of the entire research community
• Develop solutions that have general applicability
• Integrate resulting technology into
  infrastructure
• Enable re-use of software, algorithms
• Integrate individual enabling technologies into a
  coherent environment

12
Creating the new infrastructure

• Prototype systems must be built and tested using
  NPACI compute, data, and networking resources
• This user community is different from traditional
  supercomputer center users
• The compute partners have assigned people to
  install and provide front-line support for the
  software under development
• Once new NPACI-developed software is sturdy
  enough, it becomes part of infrastructure
• Work with vendors to provide similar functionality

13
Partners are of several types
14
Evolution of partnership

• Even in its first year, the composition of
  partnership has evolved
• International affiliates add expertise and
  resources in selected areas
• In each case, specific activities form the basis
  of the affiliation
• Several sites changed from associate to
  development partners due to
• Movement of people
• Identification of groups that strengthen or
  broaden thrusts

15
NPACI is a real partnership

• All projects involve multiple NPACI partners
• Executive Committee actively manages and
  allocates resources
• Reviews all thrusts continuously
• Selects activities to be initiated, redirected,
  terminated
• Members are multi-site, multidisciplinary
• 10 partner institutions represented by its 16
  members