Full HTML for
Given by Geoffrey C. Fox at Ohio supercomputer Center workshop on January 24 1997. Foils prepared January 26 1997
Outside Index
Summary of Material
| JISC = Part of UK Government supporting "Internet" (JANET) and related Infrastructure (not research) programs |
| HE = Higher Education, NTI = New Technologies Initiative |
| JISC NTI Program ran from mid 93 to mid 96 and included 8 HPCTE (High Performance Computing Training and Education) Programs and 6 cluster deployment projects (major follow up project in latter area) |
| I reviewed Program for a week in December 96 |
| Lessons are generally applicable |
| Quite a lot of excellent HPCC Course Material Produced although not presented terribly well on the Web |
Denote Foils where Image Critical
Denote Foils where HTML is sufficient
Remarks on HPCC and Computational Science Education in UK Ohio State Supercomputer Center January 24 1997
An Analysis of HPCC Education Initiative in United Kingdom
8 UK Training and Education Projects - I
8 UK Training and Education Projects - II
6 Projects in Deployment of Clustered Computing Systems - I
6 Projects in Deployment of Clustered Computing Systems - II
Teach Computational Science -- not HPCC!
Clusters of PC's/Workstations are Only HPCC Deployment in Most Places!
Don't Hide your Work in Postscript!
Long Term Commitment Necessary!
The Research Aspects of NTI was Largely Unsuccessful!
Some of the Teams were Imbalanced in Research v Infrastructure Expertise
Outside Index
Summary of Material
HTML version of Basic Foils prepared January 26 1997 
Full HTML Index
| Geoffrey Fox NPAC Syracuse University |
| 111 College Place |
| Syracuse NY 13244-4100 |
HTML version of Basic Foils prepared January 26 1997 
Full HTML Index
| JISC = Part of UK Government supporting "Internet" (JANET) and related Infrastructure (not research) programs |
| HE = Higher Education, NTI = New Technologies Initiative |
| JISC NTI Program ran from mid 93 to mid 96 and included 8 HPCTE (High Performance Computing Training and Education) Programs and 6 cluster deployment projects (major follow up project in latter area) |
| I reviewed Program for a week in December 96 |
| Lessons are generally applicable |
| Quite a lot of excellent HPCC Course Material Produced although not presented terribly well on the Web |
HTML version of Basic Foils prepared January 26 1997
Full HTML Index
HPC Training and Education Centre - Queen's University Belfast (DENI funded)
|
HPTCE: Training and Education Centre - University of Wales College of Cardiff
|
HPCTE: HPC Training and Education Centre University of Edinburgh
|
HPCTE: HPC Training and Education Centre - SEL(London)-HPC Consortium
|
HTML version of Basic Foils prepared January 26 1997
Full HTML Index
HPCTE: HPC Training and Education Centre - University of Manchester
|
HPTCE: HPC Training and Education Centre - University of Southampton
|
Use of Fortran90 and HPF - University of Liverpool
|
Parallel Computing in Higher Education - University of Oxford
|
HTML version of Basic Foils prepared January 26 1997
Full HTML Index
CONDOR: A Dynamic Self-Configuring Distributed Computing Facility - University of Durham
|
DQS: HPC-Alpha Workstation Farm - University of East Anglia
|
LSF for CFD: HPC using Spare Capacity on a Network of Workstations - University of Glasgow
|
HTML version of Basic Foils prepared January 26 1997
Full HTML Index
PVM on Linux PC's:Development of Parallel Finite Difference Time Domain Application - Brunel University
|
NQS: Support of Distributed Batch Systems for UNIX - University of Sheffield
|
| Codine: Serial Work and Parallel Computing Techniques in a Workstation Cluster Environment - Imperial College of Science, Technology and Medicine |
HTML version of Basic Foils prepared January 26 1997
Full HTML Index
| Integration of Computation into the main stream of HE curricula |
| The goal of outreach to a broad range of institutions and a broad range of subjects can be accomplished with a more general theme than the original HPCTE mandate – namely the integration of modern computational techniques and technology into education and research. |
| The up-front investment will be most fruitful if aimed at upgrading the basic fabric of the HE (Higher Education) enterprise (faculty enhancement, courses with computational modules and specialized "centres" with a computational flavor) which will be continued and enhanced in a natural way after the JISC activity is finished. |
| The most successful HPCTE centres implicitly adopted this approach while those taking a narrower view of their charter struggled in their broad outreach activities. |
HTML version of Basic Foils prepared January 26 1997
Full HTML Index
| Suggest Coordinated effort to aid accelerated deployment of low cost clusters into HE institutions. |
| JISC was farsighted in identifying importance of cluster computing but could usefully do more to capitalize its investment in terms of a pro-active effort to aid deployment of this technology (especially low cost PC's) broadly. |
| This could involve coordinated CHEST purchases (site licenses) and should deliberately focus on just a few of software and hardware choices. |
HTML version of Basic Foils prepared January 26 1997
Full HTML Index
| Encouragement of active links with related efforts outside the UK and careful review of Intellectual Property issues and tradeoff between cooperative open dissemination and protection of investment. |
| The UK activities would benefit from active collaboration with related international efforts such as the NHSE(National High Performance Software Exchange) and supercomputer centres in the US. |
| Currently the UK projects seem unclear as to whether to encourage broad international dissemination of their work or whether to protect the JISC investment by making electronic access to their material inconvenient or impossible. |
| My prejudice is that in rapidly changing fields, one benefits from open and indeed pro-active dissemination using interactive HTML and not archival ftp/postscript. However at least a consistent corporate set of ground rules should be agreed. |
HTML version of Basic Foils prepared January 26 1997
Full HTML Index
| Stronger incentives for institutional commitment to sustain efforts jump started by JISC |
| I was surprised that some excellent JISC projects appeared to be struggling to cope with "Life after NTI" and appeared to be getting little sustaining support from their host institution. This issue appears to be important in evaluating future proposals. |
| There was very modest continuing funding -- my report tried to make case to continue at lower level |
HTML version of Basic Foils prepared January 26 1997
Full HTML Index
| Suggest Stringent review of future systems software development projects which tended to be the least successful part of existing NTI programme |
| One third of the projects reviewed were developing some sort of systems (as opposed to application) software. |
| These were clearly less successful as a group than training or deployment (of other commercial or university systems) projects. |
| I didn't list these projects! -- Portable HPF from Southamptoon was successful (being continued at NPAC) |
| JISC does not have the funding level or research mandate to pursue software development projects and should limit and very carefully review its investment in this area. |
HTML version of Basic Foils prepared January 26 1997
Full HTML Index
| Careful review of qualifications of teams to ensure they have appropriate mix of both professionals and researchers at leading edge of HPC research. |
| Whether one adopts either a narrow or broad view of HPC, the field is undoubtedly changing very rapidly with critical new developments both in industry and academia. |
| Even pure deployment or training programmes need to be cognizant of the latest developments and some of NTI activities were handicapped by lacking such expertise on the implementation team. |
| Further some of the other projects were largely research based and lacked necessary professional infrastructure development expertise. |
| I recommend that JISC carefully review overall qualifications of its proposal teams. |