The Guardian technology pages today carried the story of HECToR, £113M-worth of supercomputer installed at the University of Edinburgh.
UK universities have a strong record in high performance computing. The ATLAS computer, at Chilton near Didcot, was the country’s fastest machine in the late 1960s and was installed at the Rutherford Laboratory as a resource for the research community – as a researcher in Oxford I was one of its users. A Cray supercomputer was installed in the University of London Computer Centre in the 1970s as a resource for the whole of the south east. And at the same time, Queen Mary College (as it was then), where I was working, hosted ICL’s pioneering Distributed Array Processor (DAP) which – if you picked the appropriate problem – could out-perform the Cray by a couple of orders of magnitude. Edinburgh’s Parallel Computing Centre (EPCC) stands in this tradition and was established in 1990.
Of course the power of these early supercomputers is not so far different from today’s desktop machines or multi-processor systems. But that’s exactly the point. Sure, these machines enabled the research community to compute problems that had, up to then, been intractable. But they also pioneered technologies which became mainstream in “ordinary” machines: pipelines, single-instruction multiple-data multiprocessing, resource coordination, programming for advanced architectures. One of the most important areas explored with the DAP was how the techniques and algorithms used for serial computers had to be completely rethought for parallel machines if their power was to be fully exploited. Think Grid Computing. It wasn’t just a case of optimising existing code!
Goodness knows why The Guardian has finally caught up; the contract for The High-End Computing Terascale Resource (HECToR’s official name) was signed nearly a year ago. ITasITis intends to be more on the ball as we bed in! The computer will be used for problems which in many cases are the 21st century equivalents of those tackled by Atlas, the Cray and the DAP: weather forecasting and its big brother, climate impact; simulations in these and other areas such as aircraft design and financial markets; high-energy particle physics; drug design; and so on. And no doubt, as computer scientists learn the tricks of exploiting its capacity, these will become the basis of future generations of “ordinary” machines too. They may be for academic use; but they are relevant to all of us.
|•||Inside the UK’s fastest machine (The Guardian Technology, 2 Jan 2008)|
|•||University Signs Contract For Supercomputer (University of Edinburgh press release, 22 Feb 2007)|
|•||Edinburgh Parallel Computing Centre (EPCC)|
|•||ICL Distributed Array Processor (Wikipedia entry, edited by me in the course of creating this note!)|
Links are provided in good faith, but we don’t take responsibility for the content or actions of third party sites which may use pop-up advertising or set cookies. Most news sites archive material after quite a short time; use the site’s search engine to find archive material.