High performance computing is clearly established as a key technology underpinning a great deal of modern work in science, engineering and medicine. As an institution Imperial College is at the forefront of the development and application of numeric modelling and simulation techniques. To be most effective, however, such work must have access to the most powerful and modern computing resources available.
The Imperial College Parallel Computing Centre, funded by HEFCE and EPSRC with support from Fujitsu, is designed to provide College workers with dedicated access to such a resource. It is equipped with a Fujitsu AP3000 distributed memory parallel machine, initially with 16 nodes but soon to be expanded to 80 nodes, linked to a Fujitsu VX vector processor, providing a heterogeneous computational resource amongst the most powerful in the UK.
These machines are available for use by all Imperial College workers and their collaborators. The Centre will provide support to enable the most effective use to be made of facilities and will itself carry out a programme of research into high performance computing methods and techniques linked to the application development activity. The remaining pages give details of the resources available, the structure of the Centre and document application and access procedures.
If there are any further queries do not hesitate to contact me or any
of the other Centre staff listed.
We look forward to
working collectively to make effective use of these facilities and to
help build a technology that is clearly going to be vitally important
to the College's future development.
Professor John Darlington
Director
Imperial College Parallel Computing Centre
September 1997
The need for an enhanced capability, dedicated to Imperial College use, was recognised and resulted in a proposal under the 1996 HEFCE Joint Research Equipment Initiative, supported by Fujitsu. This was successful and resulted in an award of £1.65M, the largest to any institution under that scheme. This has funded the installation of a configuration consisting of a Fujitsu AP3000 massively parallel processor, to be expanded to 80 nodes, linked to a Fujitsu VX vector processor, providing a theoretical aggregate peak performance of nearly 50 GFlop/s with 15 GBytes of memory, 364GBytes of local disks and more than 400GBytes of available high speed disk storage.
These machines are available for use by all Imperial College workers and their collaborators. The Centre is able to provide assistance to application developers to enable them to make effective use of the facilities in developing innovative applications and, within the Department of Computing, will be linked to research in high performance software methods and environments and numeric techniques.
The sophistication and range of applications for this technology is rapidly expanding. The Centre's role is to work collectively with members of the College to take maximum advantage of all the opportunities available.