HPCN in Europe - A Critical Assessment

Bristol, 03 December 98 There is a crisis in HPCN. During the recent years, the area has become a marginal one in the IT sector. It is no longer the cutting-edge-technology-area that attracts the brightest young people. There is no real progress in parallel computing and HPC systems, supercomputers, are just one option in a large growing catalogue. In Europe, the crisis is even worse: there is no HPC industry, no HPC community, and no European HPC policy. When Europeans active in HPCN meet, it is in the USA. On the positive side, Europe is still a large consumer of HPC technology from the USA and Japan and applications that use a little bit of HPC technology are quit successful.

The HPC hardware industry has almost ceased to exist. The lasts of the Mohikans are British/Italian QSW, Scali/Dolphin in Norway and Supercomputing Systems in Switzerland. They are not really supercomputer companies, but rather network interconnect and integration manufacturers. The companies lack political and financial support from an Asci like European programme and are to small to survive in the long run as independent hardware producing companies. Perhaps in close co-operation with or as part of a large companies like Siemens (in the case of Scali) or Compaq (for QSW and Supercomputing Systems) they can exist for a longer period.

One of the problems of the European hardware companies is their visibility: nobody knows them: they hardly publish press releases and in one case have a web site that has not been updated for over two years. This lack of publicity policy is a marketing problem typical for many European IT companies: still they have the idea that good products do not need advertising.

The major US and Japanese supercomputer companies have become active in Europe and sell machines over here. Analysing the TOP500 list, one can conclude Europeans do not have a real preference for one or the other: they buy, percentage wise, as many American and Japanese supercomputers as is happening on a world wide scale.

There are hundreds of HPCN and middle-tier software tools originating from Europe. They are mainly developed and used in research environments. The EuroTools initiative provides a - partial - overview. European HPCN software companies are not numerous: there are only a few, not enough to really call it a HPCN Software Industry. They probably will not survive as HPCN tools companies, but could if they diversify into other sectors.

In applications, Europe is still rather strong. From a number of applications optimised versions for HPC equipment are available. They sell, however, most of their licenses for other type of equipment. ESI, with PamCrash and NAG with their numerical software, are some examples.

It is only in the usage of HPCN technology that Europe - at least part of it - is doing as well as the USA and Japan. In the high end, the supercomputer centres still play an important role. There is a trend in which part of these centres evolve in concentrations of equipment, as for instance Amsterdam and the large German supercomputer centres, and others evolve more into technology transfer centres that are drifting away from HPCN.

In industry, there are many, medium sized and small supercomputers and in general there is enough knowledge to use HPCN technology competitively.

European HPCN community

Apart from users, hardware and software producing companies, a community needs several other element that act as a glue to keep the community together. Is there enough of this glue to keep a Europen HPCN community together?

The single most important European contribution to the (world-wide) HPCN community is the TOP500 list of the world's most powerful supercomputers, that originated in Mannheim. It is widely accepted as "the list" and used by vendors and users.

The Primeur HPCN service is another glue item: it has a large audience in Europe and is the only vehicle for European news. Primeur has a weekly edition with short news flashes in e-mail and fax version, monthly editions, live special issues, analysis and resource centre links.

HPCnet, a network of HPCN developers and users, which did exist for three years, was a society type of organisation with good workshops and a comprehensive web site. However, HPCnet stopped when the funding from the European Commission stopped.

The TTN network is an expensive effort to get "HPCN technology" into industry. However, it is not yet clear whether this will have any significant success.

In several countries there are local HPCN organisations. Examples are ORAP in France, Speedup in Switzerland and the Platform HPCN in The Netherlands.

On a European scale, there are only academic events that have some significance: examples are EuroPar and HPCN Europe. For industry, there is the Mannheim Supercomputer event that is more German oriented and to some extend, the RCI meetings.

Do we need HPCN?

In most reports a few years ago, HPCN was seen as an enabling technology, that is vital to the modern IT based industry. Within FrameWork IV of the R&D programme of the European Commission, there was a special programme line for HPCN. This has, however, not resulted in self-sustaining hardware or software industry with a significant size. In the Framework V programme that starts this year, HPCN is not mentioned anymore.

Belgium is a country that economically is doing just as good as other European countries. Flanders, for instance, is wealthier that the Netherlands. The country, however, has never had a significant TOP500 presence ever, and it had no national support programme for HPCN during the recent years. It is strong in several other areas, like speech-technology, so it is not that they do not know how to produce technology. So do we really need HPCN?

Perhaps, we should abandon terms like HPCC and HPCN and go back to call it supercomputing, what it is after all. Then the market in Europe is mainly scientific/ technical and there are some 5000 people active. Not very large, but not insignificant either.

Ad Emmen