Let\'s keep it simple this time!!

I\'ve had another crash, of a different model this time at about 75%. It\'s on my 4 core machine. Computer ID 532553 I think, according to the messages page. I have uploaded it this time, and have already restored it and run again from a back up one. I think it crashed at the same point. Could you find the reported result and tell me if it is a no hoper again. Where would I look myself to find the \'negative\' messages that tell you it has gone outside of the paraameters, so I don\'t have to bug you guys so much with future failures. Thanks guys, and no one mention 2050 this time!!!!! ;-)

Cheers.

Pete McCann

The \"Negative ...\" messages are in the list of error messages on the server page for the model in question.
Just go to your account page, (LOTS of ways to do this), click on \"View\" to the right of Results, and then on the Result ID for the model you want to examine.

The page for the model you\'re talking about is probably this one.

Hi Peter

Bad news I\'m afraid:

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=6369447

Negative pressure again. This means as you thought that the initial parameters were not viable, but the only way to discover this is to run the model.

Very very occcasionally the problem is caused not by the parameters but by a computation glitch on the computer, in which case a restore of backup should enable the model to advance beyond the previous crash point. If the problem is the parameters, as is almost always the case, the restored model will crash again at the same point.

Let\'s hope your other 3 models have better sets of parameters (though scientifically they\'re all useful).

You can get to the results yourself by opening the Tasks tab in boinc manager and clicking on Your results. Go to the list of computers, then the list results for the computer you choose. You can then look at graphs, trickles, errors for each of the models.

The web page for each model only ever shows the error messages for the FIRST crash reported. This means that if a model has crashed and been restored, whether it subsequently crashes again or completes, the first error is still the only one recorded there. But if the model completes, the server is under VERY STRICT INSTRUCTIONS TO ACCEPT IT NEVERTHELESS.

Hope that helps.

Edit - Les was quicker off the mark!

---

Cpdn news

I seem to recall that AstroWX had something like 5 consecutive NEGATIVE THINGY crashes?, it\'s quite a high probability outcome.

But at the end of the day, the model has run to the end of it\'s natural life, and produced all the scientific data that it can. So you shouldn\'t feel bad about it.

Since I\'m not allowed to mention 2050, I\'ll mention 2040 instead :-)

Your result reached 2041, which is past the important 2040 milestone. Whenever a model reaches a 40 year milestone (1960, 2000, and 2040), it uploads extra information to the server (you may have noticed that these uploads are bigger than normal).

PS Great computer that, I\'m jealous :-)

---

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

Since I\'m not allowed to mention 2050, I\'ll mention 2040 instead :-)

Your result reached 2041, which is past the important 2040 milestone. Whenever a model reaches a 40 year milestone (1960, 2000, and 2040), it uploads extra information to the server (you may have noticed that these uploads are bigger than normal).

PS Great computer that, I\'m jealous :-)

Hi guys.

Thank you all for the input. Ah well. That\'s life again. At least we did make it to 2040 (though 2050 would have been better!) Ssssshhhhh. Don\'t say that!
:-)

As for the computer, I built this from bits bought off Ebay for about £300. Not bad hey? I\'d like a pair of opteron 285\'s to put in it though! That would really put some Horse Power under the bonnet. Still all 4 models were chugging along at about 2.6 s/timestep, so it is a good work horse, and increadible for the money.

Mike, AFAICR, it was seven spread over five machines. (I haven\'t the heart to go through to record to verify my memory!) Unfortunately, two Models from that era remain in my mix, one on each of two boxes; my hopes are not high for them. (Prolonging the \"anticipation\", one is suspended while the E6600 box runs Beta.)

Our consolation, Pete, is that those crashes aren\'t failures in the ordinary sense of the word. The science team receives confirmation about the behavior of particular parameter mixes when associated with particular oceans.

As an aside, these Models are, in a real sense, 360-year experiments; the part we run now is the last 160 years. (A group of us ran 61 200-year \"Spinup\" Coupled Models before release of the Coupled Model publicly, here and for the BBC-supported side of the Project. They are identified in the Run\'s parameter list [Ocean Start Dump and Ocean Flux Correction; the number identifies the particular Spinup Run]. The Oxford staff [rightly!] concluded that asking everyone to run the full 360-year experiment would be a task too far.)

We should have fewer Negative Pressure/Theta results with the current crop of Models. (He says, hopefully.)

[Edited for typo.]

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.