Task hadsm3fub_013j_005927676_0 using hadsm3 version 506

The task is at 85.941% for a week now. It goes a little higher to 85.955%, then goes back to 85.941%. It appears to be an ice world on graphics, all blue. Timestep 149910 of 259248,Date 04/08/2059 3:00 (1810 to 2050)

Do I need to abort this task.

Thanks

All that is necessary for the triumph of evil is that good men do nothing.

---

Hi Gordon, and welcome to the message board.

All the models in that work unit have hit the same problem, though not everyone has noticed. The model appears to change into an ice world between 140,426 and 151,228. The other models aren\'t looping indefinitely, but are making very slow progress - about a week per trickle. They will eventually finish, but that\'s about ten weeks - in which time the PC could possibly do four other slab models.

So, I would abort it.

Iain

Iain, if you think it\'s in order, I\'ll send private messages to the other people running the same workunit, though I\'ll wait till each person\'s model reaches the critical point and their timesteps & trickles show they\'ve hit the same problem. When we notice a problem it seems perverse to let crunchers waste computer time on a doomed model. What do you think?

---

Cpdn news

Iain, if you think it\'s in order, I\'ll send private messages to the other people running the same workunit, though I\'ll wait till each person\'s model reaches the critical point and their timesteps & trickles show they\'ve hit the same problem. When we notice a problem it seems perverse to let crunchers waste computer time on a doomed model. What do you think?

They may be doing it deliberately, but I doubt it. When I first got an iceworld I thought I would try to finish it, as the slab Zips don\'t get uploaded until the end of the phase - but I didn\'t have the patience! I convinced myself that if the project really wanted that kind of data they would re-write the model ...

The bad news is that about one in seven of my slabs has gone awry somehow: if that\'s applied across the whole project, then that\'s a lot of PMs - though, as you say, it could be limited to the work units of people who pitch up here with this problem. Caveat cruncher?

I didn\'t mean I am going to trawl through every member\'s models. For one thing, the server status page says there are over three quarters of a million CPDN models in progress(!!)

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/server_status.php

I just meant the people running the same model as Gordon because now we already know about them we may as well make use of the knowledge. And when other people post about a similar problem we could spend a moment looking at the trickles of other members running the same WU then if necessary tell them about their problem. If some of these people are running BOINC as a service with no graphics they\'re unlikely to notice the anomalies.

---

Cpdn news

Sure. I was only being awkward.

One thing to watch out for is that an iceworld may be limited to a processor/operating-system combination - so, Intel/Windows may freeze, but AMD/Windows not. The best check is to look for a significant increase in the S/TS on the same combination and at the same timestep as the person who has spotted the problem - as has occurred with Gordon\'s work unit.

29 Feb 2008 12:58:57 814783 7220684 hadsm3fub_013j_005927676_9 3 183,634 2,275,557 3.2409
21 Feb 2008 21:11:33 814783 7220684 hadsm3fub_013j_005927676_9 3 172,832 1,786,674 2.5844
16 Feb 2008 04:51:06 814783 7220684 hadsm3fub_013j_005927676_9 3 162,030 1,297,841 1.9071
10 Feb 2008 10:10:37 814783 7220684 hadsm3fub_013j_005927676_9 3 151,228 808,730 1.2076
09 Feb 2008 14:13:54 814783 7220684 hadsm3fub_013j_005927676_9 3 140,426 738,815 1.1212

and

23 Feb 2008 02:06:23 830206 7220682 hadsm3fub_013j_005927676_7 3 151,228 1,505,392 2.2478
20 Feb 2008 00:45:04 830206 7220682 hadsm3fub_013j_005927676_7 3 140,426 1,378,477 2.0920

Thus, Chris Beaugrand and GOAL: Mexico\'s 1st place should get a heads-up.

PS Gordon has aborted the model.

Yes, they were the ones I was thinking of sending a PM to ie where the problem has already shown up. Then wait to see what happens with the others (assuming one remembers to look back a week later). There\'s no point in telling people about a potential problem that may not occur on their computer.

---

Cpdn news

Shall I do Chris and you the potential Spanish speaker?

Good idea.

---

Cpdn news