Alan Sugar and the secrets of the universe
Published: 08 May 2007 16:20 BST
In 1984, UK home computer purchasers had a wide range of options. For around £10m you could buy one of the world's fastest computers, the Cray XMP. For a few pounds less, you could buy one of the world's slowest — Amstrad's new CPC 464 clocked in at £300. Alan Sugar, then as now Amstrad's ferociously populist head and the man responsible for more cheap CB radios, wonky audio centres and clanky video recorders than anyone outside the Far East, was as far removed from Seymour Cray as Colonel Sanders is from Marco Pierre White.
He still is. So his appearance to open Queen Mary, University of London's new scientific computing cluster is yet another pointer to the speed at which distributed computing is changing the high-performance landscape. Ninety percent of the cluster is built from servers from Amstrad's Viglen subsidiary, with the latest hardware bringing 280 dual-processor dual-core AMD machines together with Linux. As well as coping with multiple tasks within the college, it will be linked up with others around the world to help provide the grid resources needed for Cern's Large Hadron Collider experiment, peeling apart the layers of fundamental physics. That's a big leap from Manic Miner.
But there are big problems in little systems, no matter how high they're piled or how cheaply they're sold. The hardware will continue to shrink, with the number of cores on a chip continuing to multiply according to Moore. Dual core is useful, quad core is interesting — but while the 80-core monsters in the lab may be keen to devour scientific data, most enterprise and consumer applications will find them as relevant as our 1984 home buyer would a Cray XMP. For most of us, Moore's Law no longer applies. As the IT industry is driven by the money spent by most of us, that's not something that will fix itself.
Researchers, designers and long-term strategists within the industry are keenly aware of this, and have been for years. But the increasingly frenetic research into making parallel processing general purpose has yet to make the breakthroughs required. There are any number of demos and what-ifs, but none that replicate the special blend of herbs and spices which made mass-market computing the defining technology of the past 20 years. Alan Sugar was cutting the ribbon at Queen Mary not because high-energy physics has suddenly become Amstrad's new market. He was there because, as yet, there's nowhere else for him to go.
Did you find this article useful?
6 out of 7 people found this useful
- Share this article:
