Goodbye

Chris,

Do you know anybody that has build a rig using expansion backplanes and is crunching 7X24 ? I am interested and intrigued to say the least - looking at my RAC you will know why I just couldn't keep things cool this summer here in Dallas with 25+ days over 100F. Thinking to unloading some machines and going this expansion route. Comments most welcome.

If the "only" reason for shutting down the computer(s) is be course of the heat, I down understand why not just throttle the CPU/GPU, then the PCs wont get too hot and your still crunching.

The idea behind S@H was to use spare cycles

I don't think it works like that any more. As soon as your computer becomes idle and BOINC starts crunching, it's going to use 100% of the CPU by default...

Actually, I think that low-power idle instructions on the CPU were not always available. Back in the 386 and 486 days, I believe free cycles were actually free--the CPU essentially ran at 100% all the time. I could be wrong on that of course.

Actually, I think that low-power idle instructions on the CPU were not always available. Back in the 386 and 486 days, I believe free cycles were actually free--the CPU essentially ran at 100% all the time. I could be wrong on that of course.

"Unused CPU cycles" meant that when you were not using your computer to run a program like a word processor or spreadsheet etc, the cpu would basically run in idle mode. This was envisaged to be in between work tasks, or at tea breaks, or lunchtimes. At that point, Boinc/Seti would kick in and use the CPU for crunching, until such time as the computer was used for another work task.

The thinking went that if the computer was powered up anyway, why not use the computing power for Boinc/Seti's benefit at appropriate times, that wouldn't interfere with the primary function of that computer.

As I've stated many times before, in many threads, the initial purpose of Seti Classic and Boinc, was to prove the viability of Distributed Computing utilising the general public. This it has done, and admirably too.

But it was never designed, scoped out, or intended, to cope with farms of dedicated computers running GPU's 24/7 like we see today. This is the 21st Century, the world has moved on, Boinc hasn't.

I can empathise with others on this thread, giving up.... It does look like there's now more crunchers, than there is work units... so maybe SETI is at max capacity now and doesn't need more crunchers...???

Just a thought..

Bob

I can empathise with others on this thread, giving up.... It does look like there's now more crunchers, than there is work units... so maybe SETI is at max capacity now and doesn't need more crunchers...???

Just a thought..

Bob

Well, that's at least what it looks like. Just now when there's months and months of shortie floodings, it's certainly true. However when the telescope is used in another way, sometime in the future, and there's a more balanced production of WU's with other AR's, I think all will settle down again.

Sten, don't go optimistic on us now!

The Arecibo schedule starting August 27 shows the ALFA receiver being put back on the telescope September 1 and the A2010 ALFALFA project's fall observation campaign starting ~03:00 UTC September 6. When some of that data gets shipped to Berkeley, it will be mainly midrange ARs.

In terms of our demand versus data availability, with no ALFA observations between July 24 and September 6 I hope there's enough in HPSS storage to keep the flow going. If not, there may be an opportunity for the project to do some local maintenance or reorganization which is difficult while in full operation.

Although there's ample data from GBT to fulfill all demand for many months, I judge the project will want to do careful testing of modified applications before starting production crunching here.

Joe

What we also need is Nitpicker to start on the 11 years of data we have already got .......