Hi,

The machines that I have running climate prediction seem to be giving less credit back than usual recently. When I checked the tasks they are running have report deadlines in some cases of 1901 and other cases of 28 Nov 2007.

These machines are stable overclocks of core processors and they run your work 100% of the time. There is no way they could not complete a task on time.

What could cause this?
Am I getting full credit for the CPU time that is being donated?

Regards,
Herb

The 1901 deadline problem is related to Boinc versions prior to 5.8.16 combined with deadlines over 999 days. More recent Boinc versions can cope with longer deadlines.

Deadlines are ignored at CPDN in any case - as long as you\'re only running climate projects, it won\'t have an effect. However, if you run other projects, you\'ll notice that the scheduling will be affected by the deadline - it\'ll give priority to any tasks whose deadlines are under thread.

Neither of these should be affecting your RAC - a sudden drop in RAC is often a hint that you might have a problem with a model - for example, a \'slow processing iceworld\' might need to be aborted. If you locate which PCs have experienced a drop in RAC, and have a look at each model running on that PC, you should be able to identify which are causing a problem.

It is worthwhile having a browse through the README posts, these can be found via the link in my signature.

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

Hi, me again.

Some more information - I upgraded the version of the boinc client a couple of weeks back, could that have affected anything?

Also I checked on the tasks that are scheduled for finishing by Nov this year and they appear to be shorter as they will complete in ~270hrs c.f. ~1150hrs for the usual tasks. And are labelled as UKMet Office not like the others which are HadCM3 ones.

Hope that helps,
Herb

---

Hi Herb,

You may want to check the graphics on this slab model on the dual core machine. It has slowed down a great deal over the last step, and may have become an \'ice world\'. If that has happened then the temperature graphic will be blue. If so, then I suggest aborting it as it\'ll take a long time to complete (there are 16 steps to go in phase 3, at maybe 4 days a step). The ice world is doomed - it\'s (probably) not caused by anything you\'ve done, and nothing you do will rescue it.

The coupled model on that machine looks OK.

Iain

Hi, me again.

Some more information - I upgraded the version of the boinc client a couple of weeks back, could that have affected anything?

Should be OK. Note that if an older version of Boinc messed up the deadlines they\'ll stay messed up after Boinc is updated. But the deadlines don\'t matter unless you run other projects.

Also I checked on the tasks that are scheduled for finishing by Nov this year and they appear to be shorter as they will complete in ~270hrs c.f. ~1150hrs for the usual tasks. And are labelled as UKMet Office not like the others which are HadCM3 ones.

There are now three different model types that can be downloaded to your PC - these are HadCM3, HadSM3, and HadAM3 (only HadSM3 can turn into an iceworld). These are discussed in more depth in the \'README - running the model\'. You are able to control which types of model download to your PC via \'your account / CPDN preferences\'.

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

Thanks Guys.

I checked the graphics and its not all blue so no ice world. I\'ll leave it running, I\'m sure it will get there in the end :-)

Good to know the report deadline isn\'t affecting anything.

Thanks for you help, have a great NY\'s,
Cheers,
Herb