I am currently running Work Unit ID 6105485 - It appears to be a blue world with no rainfall and no pressure, in fact there is nothing to see on the graphics of this unit!! It is also strange as it seems to be in the 1800s (currenntly 18/04/1814). It is also going a lot faster than the previous work unit (about 10 times faster in percentage terms). Should I abort?

I am gutted that my previous model crashed out!!

Steve

Hi Steve,

Yes, I would abort it. It appears to be a \"fast processing iceball\" and will not produce a viable or useful climate. The fact that this model started in 1811 is due to it being a different type of model than the others you have run previously. See http://www.climateprediction.net/science/strategy.php for more details on the different model strategies.

Thanks for the quick response. I will abort.

Hi Steve

I\'ve had a look at your previous models to find clues about how you can avoid future model crashes:

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/results.php?hostid=635614

The last one crashed with a 22 error, which unfortunately could be either a fault in the model or something that went wrong on your computer. But your first 2 models crashed with -107 errors which do help with a diagnosis. These tend to be graphics errors.

I think you should look at the Project READMEs which you can get to through my signature. In the README about Solutions to models crashing, look at item #5 by Mike. It contains advice on the precautions required to avoid -107 and the other common types of model crash. Within that post you\'ll find a link to Thyme Lawn\'s post about updating graphics card drivers, which is something you should do. This will be a free download from the web.

Hope that helps.

---

Cpdn news

I have had some of those blue worlds. Every time I have restored from backups and the models have finished in the end, but I\'ve had to restore 2-3 times.
regular back-ups is a must for the project cause a lot of models fail, especially the longest ones.
Even though the models generate valuable data even if they dont finish, finishin the models to the end should be the primary goal.

I agree totally. Completed models are much more useful for the researchers and more satisfying for their crunchers. Even the most experienced crunchers can have a model crash (Carl, our ex-programmer, once crashed a model on his home computer!) so regular backups are our defence against disaster and disappointment.

---

Cpdn news

[quote]Hi Steve,

hello everyone I\'m wang haosu I am from CHINA

---

Hi Wang Haosu

Welcome to ClimatePrediction. I see you have two new models. Good luck running them. If you need any advice about processing them, just post on the forum.

Mo

---

Cpdn news