HECToR launch : another free lunch ?

January 14, 2008

Well, in today’s comments someone accused me of being on an ego trip cos I was brown nosing with Mason, Wade, Efstathiou, Rowan-Robinson et al at the “ad hoc” meeting. So maybe I shouldn’t mention having a free lunch today, listening to the Chancellor of the Exchequer tell us that Science is safe in his hands, and chatting to O’Nions over a glass of champagne ? Except of course I didn’t get within twelve feet of that nice Mr Darling, and O’Nions pretended to know who I was but actually couldn’t quite remember. I only mention this to point out the contrast with the ad hoc meeting… everybody at the ad hoc meeting was someone I’ve known for twenty years or more, and have sunk many pints with, so it didn’t exactly feel like an ego trip… Old Boys Club, maybe …

So why was I hobnobbing with the big boys today ? It was the official opening of HECToR, the new national supercomputer, paid for by EPSRC but run by the University of Edinburgh. Its the seventeenth most powerful computer in the world apparently. If you like lists, check out www.top500.org. Its a Cray XT5, and I finally got to see the beast today. Rows of black boxes with identical racks of identical boards. One of the Crayons showed us the inside of a board. This was weird cos it looked just like the inside of my PC. Opteron CPUs and a bunch of DIMMS. Oh and some super fast interconnect ..

Meanwhile, the IOP and RAS have released their joint submission to the upcoming select committee. You can find it here.


Gemini and STFC’s problems : triple squeeze

November 25, 2007

Michael Rowan-Robinson, President of the Royal Astronomical Society, has written a piece for Research Fortnight about the Gemini-withdrawal issue. The RAS has put out the text of Michael’s article as a newsfeed. Its worth a read. I think Michael has the issues spot on (he usually does..)

STFC has a triple squeeze. Because of “Full Economic Costing” they have to give the Universities more money; the subscriptions they have to pay to CERN, ESA, and ESO are tied to GDP which is going up in real terms; and the operations costs of Diamond and ISIS are alarmingly larger than forecast.

UK astronomers understood that moving in with the big boys was going to be scary. When Diamond sneezes, we catch cold. But if something folds, we could be in the gravy. Hmm. Sorry about the gharrssly mixed metaphors.

Key point buried in Michael’s article : watch the spreadsheets at your University and make sure you are getting the FEC flowthrough. The biologists will be getting nervous as Charities, their main grant sources, don’t pay FEC, so their income generation now looks much weaker…

Details for political geeks only :

I was aware of the ops costs issue in general terms, but Michael refers to a report of the Public Accounts Committee which spells this out. You can find the report (HC 521, Nov 13) here. Dull reading but important … Big projects like Diamond, ISIS, MICE, and HECToR have all been very impressive in capital terms, mostly coming in on time and on budget, but they are nearly all coming in 50-80% higher in operations costs terms than originally approved. This particularly hits STFC, and looks like costing them £27M/year, even before paying all that FEC and growing subscription costs.

EPSRC will be hit too, as the new national supercomputer, HECToR has, like Diamond, ISIS-2, and MICE, come in on budget in capital terms, but has turned out to be more expensive to operate than originally expected. This is where I breathe nervously. HECToR is operated by EPCC within my own School. Its in an anonymous building out near where Dolly the Sheep was born. The costs of HECToR are dominated by the huge electricity bills, needed for cooling the darn thing. We are currently trying to work out if we can vent some of the heat into greenhouses and grow tomatoes. I jest not. Anyway, with the oil age about to end, electricity ain’t going to get any cheaper.

However … the contract we have signed makes it clear that EPSRC bears the operating costs, not the University of Edinburgh.

Or if thats not true, Richard Kenway and Arthur Trew are dead men.