This page contains a Flash digital edition of a book.
HPC APPLICATIONS: DESKSIDE HPC


Economies of scale


Stephen Mounsey examines the role of deskside supercomputers, and looks at some of the different ways of bridging the gap between a $200 workstation and a $200m data centre


H


igh-performance computing is a powerful resource, to say the least. Where a well-designed system has been intelligently employed, HPC has been the driving force behind a huge number of cutting-edge operations, innovations, and discoveries. No matter what else it is, however, HPC does not usually come cheap. World-leading HPC installations naturally present world-leading costs – the $208m price tag on the Blue Waters petascale computer, for example – but even a modest cluster is a significant investment beyond the reach of many would-be users of HPC, particularly in the category of small- to medium-sized enterprise. Now small but powerful supercomputers are making HPC available to many customers for the first time.


Barry Bolding, director of product marketing at Cray, explains how so-called deskside (or desktop) supercomputers fill a gap in the market: ‘A low-end workstation has traditionally consisted of one or two processors or “sockets” as we like to call them. Today, a single socket may have four or eight cores in it, and that’s a great offering for somebody doing word processing, code development, and other tasks with low computing requirements,’ he says. Bolding notes, however, that Cray’s expertise is in computationally demanding simulation and modelling applications: ‘This is typically a scientific arena, although today, other areas such as finance, digital media, and the movie industry are also doing simulation and modelling tasks. Typically, computing requirements for the simulation and modelling applications start at around 10 times the performance


32


Deskside supercomputing offers a middle-ground for users who need more than a workstation can offer, but don’t need a full cluster or grid solution


of a typical personal computer, going up ad infinitum to millions of times that level of performance.’


According to Bolding, Cray aims to cater for developers of modelling and simulation applications who are at the earlier stages of development, when they are just beginning to develop the models, or when the models are very small. ‘That is the deskside supercomputing space. We feel that it’s really an area for those developers, and for those modellers,’ he says. ‘If you look at something like the automotive or aerospace environments, or life sciences, you typically have people testing out the first stage of their models, or testing out new parameters. At that stage they don’t really want to be sharing a giant machine, waiting three or four days for their simulation to be done, using millions


SCIENTIFIC COMPUTING WORLD OCTOBER/NOVEMBER 2010


of CPU hours. They want to be able to get a quick turnaround, making sure that the models are correct,’ he says, adding that the flexibility of deskside HPC makes it appealing to such customers. The other customer group for deskside HPC, he says, is that of company or university researchers whose computing requirements never quite reach a full-blown supercomputer, but are still much greater than a workstation can provide. ‘They need a machine of sufficient specification to be able to handle a complex simulation model,’ says Bolding. Christian Tanasescu, vice president of software engineering at supercomputing firm SGI, reiterates the importance of flexibility to the mid-level user of HPC: ‘The market need for a personal supercomputer comes from the general trend towards lowering the barriers to


www.scientific-computing.com


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48  |  Page 49  |  Page 50  |  Page 51  |  Page 52  |  Page 53  |  Page 54  |  Page 55  |  Page 56
Produced with Yudu - www.yudu.com