Hi,

My one computer has a problem for two days where all the models it downloads are ended with an error. It downloads a model and decides it cannot calculate the model. After four models the server does not give any new models to the computer.

Is this a problem in the model ? If so than the \'daily quota\' should be higher because the computer is now dorment all day. When checking the results it seems that more computers have come up with the same problem.

gr.

Ronald

Have you restarted your computer recently? With windows a periodic restart can clear up some issues.

Is the client giving you an error message about being unable to delete a file? Sometimes the OS will lock a file that needs to be deleted. If this is the case exit the client and manually delete the file. A file that cannot be deleted automatically can cause all workunits downloaded to error out.

Having a low quota is a good idea for this project since each workunit that errors out may wait up to a year before it is resent. If the workunits are erroring out after a few seconds thousands can be run through in a fairly short period of time.
<br>John Keck
BOINCing since 2002/12/08

Hi, a model takes 3 weeks to run even on the fastest computers, so there is a problem with the model crashing on your computer, hence I limit it to 4 per day (before this some people had gone through thousands of workunits per day)!

If your host repeatedly crashes workunits on CPDN, especially early on in phase 1 or so, I'm afraid your computer just isn't compatible with the project (the climate model is "sensitive" in more ways than one), and it would probably be best to detach and attach to another BOINC project.

&gt; Hi, a model takes 3 weeks to run even on the fastest computers, so there is a
&gt; problem with the model crashing on your computer, hence I limit it to 4 per
&gt; day (before this some people had gone through thousands of workunits per
&gt; day)!
&gt;
&gt; If your host repeatedly crashes workunits on CPDN, especially early on in
&gt; phase 1 or so, I'm afraid your computer just isn't compatible with the project
&gt; (the climate model is "sensitive" in more ways than one), and it would
&gt; probably be best to detach and attach to another BOINC project.
&gt;
&gt;

I see the need for the maximum downloads per day. However when I look at the results I see that other computers also crashed on these models, so I presume it's not only my computer. Besides that I think an AMD64-512mb should be able to handle the models. I was hoping to resolve the issue, but the advice seems to be to leave this project and start another Boinc project (wich I did).

regards,

Ronald

hi, it's not a question of your computer "can't handle" (i.e. not "powerful"), it's just that there are some configurations out there that just don't like this huge fortran climate model. I think I have made some changes (esp in the "-5 error") this weekend that I hope to get out for the launch that should help solve some of these problems; but I have seen incredible machines that weren't overclocked or anything just refuse to run this. I hope when the MetOffice buys supercomputers they get a money-back guarantee if it doesn't run! ;-)

@Carl

No problem. I just figured that Beta testing would want to know about the problem and maybe wonder why this would happen. The answer 'detach and find another project' suprised me especially since this is, for me, a normal computer. Maybe even the best 'test computer' since it has only windows and boinc on there (it acts as a server)! So in the back of my mind I am wondering why this computer wouldn't like this huge fortran model and why a smaller more busy computer would.

But it's not a problem. I will be happy to use the computer for another project.

I think some workunits were sent out with bad parameters that may have caused these crashes, so perhaps it is explainable after all. I am going to find out more tomorrow.