Hi,

I've been running 3 projects on BOINC, and only have questions here.

Running Win98SE, BOINC 4.25 on a 1.5 processor clocked at 1.7 at the store.

The climate project occasionally pops up an error box, loaded with UNKNOWNs, and then goes on.

The projected time to completion jumps and falls alternately, and the disk space has jumped from about 55MB when I first noticed it, to 110 now.
(lots of space left - GBs)

First question: are the error boxes causing problems?
Second: will the disk usage keep increasing?

I guess overall, my question is: Is this normal?
Is there a place where one could find a normal storyline on this project?

Ralph

Hi Ralph
Welcome to the project.

Problems with Win98 were only fixed about last October. They have reoccurred recently with the new, 'signed',
versions of hadsm, V4.10 on. Why, and when/if this will be fixed again, is unknown. The programming effort at
the moment is going to the newer, more complex phases of the experiment. This probably won't change until a
second programmer is appointed.

If a model keeps running, OK. If not, you may be out of luck. For better help, copy and paste the error messages here.

My twin models take about 5 weeks/700 hours to complete on a P4 3.2G machine.
Disk space:
about 700Megs per processor. (So 1,400M for HT machines).
Plus about 330Megs per completed model, in the form of lots of zipped files.
Plus some for the programs.


Normal? I don't think we have a 'normal' here. :)
But you could try the FAQ to the left of this message. And the Climate Science pages, (also to the left), if you want more on why this is such a large, slow project.
And there is the Climateprediction.net discussion forum, where there lots of discussions (what else?) on all manner of subjects.

Les

Thanks Les,

Well, I got my answer a little later in the night.

"04/14/2005 07:05:12 PM|climateprediction.net|Unrecoverable error for result 1nu7_000098346_0 ( - exit code -5 (0xfffffffb))"

I'll try wading through the FAQ while this next attempt runs.
Do the failed attempts help any?
I appreciate the humour, having trained in programming.

Thanks again for the prompt reply!

Ralph


> Hi Ralph
> Welcome to the project.
>
> Problems with Win98 were only fixed about last October. They have reoccurred
> recently with the new, 'signed',
> versions of hadsm, V4.10 on. Why, and when/if this will be fixed again, is
> unknown. The programming effort at
> the moment is going to the newer, more complex phases of the experiment. This
> probably won't change until a
> second programmer is appointed.
>
> If a model keeps running, OK. If not, you may be out of luck. For better help,
> copy and paste the error messages here.
>
> My twin models take about 5 weeks/700 hours to complete on a P4 3.2G machine.
> Disk space:
> about 700Megs per processor. (So 1,400M for HT machines).
> Plus about 330Megs per completed model, in the form of lots of zipped files.
> Plus some for the programs.
>
>
> Normal? I don't think we have a 'normal' here. :)
> But you could try the FAQ to the left of this message. And the Climate Science
> pages, (also to the left), if you want more on why this is such a large, slow
> project.
> And there is the Climateprediction.net discussion forum, where there lots of
> discussions (what else?) on all manner of subjects.
>
> Les

---

If I get good service, I tell 10 friends;
if I get bad service, I tell 10 friends

Hi ralph
The -5 error is, apparently, an all purpose one. One of problems labeled with it, is a negative pressure calculation
in one of the many cells into which the atmosphere is divided. A message to this effect is written into one of the
files in the failed model. I had one soon after starting, but have never gone back to look.

All I can suggest is keep trying.

More humor: I see you have AMD processors. The current fortran compiler is an Intel one, which favors Intel
processors over AMD in terms of speed. And the joke is: The problem is "No Intel Inside".

One wonders though, if it also doesn't handle some instructions properly in AMD processors, as a lot of users with
problems have one. Latest rumor is that the programmer has gotten hold of a 64bit compiler.
Watch this space. Or rather, <a href="http://www.climateprediction.net/board/viewtopic.php?t=2858"> THIS</a> space!

Les

Hi again,

Well, I seem to be still running. (?)
The client contacted the server, and now my disk usage is up to 140 MB.
I'm not sure if I'm on different data, or continuing on.
OK, the .zip file numbers are different, so I'm on to another one, right?

I know the problems of processor differences from way back in the '80's.
In the local programming group for Texas Instruments, we had some very
talented local people manage to cope with the three different processor models
in the pseudo-assembly of the time.

I'm looking at upgrading in the foreseeable future, so I'll take the
opportunity to check the different processors. This one WAS a bargain
3 years ago...

Thanks again; I'll keep pluggin' away!

Ralph

&gt; Hi ralph
&gt; The -5 error is, apparently, an all purpose one. One of problems labeled with
&gt; it, is a negative pressure calculation
&gt; in one of the many cells into which the atmosphere is divided. A message to
&gt; this effect is written into one of the
&gt; files in the failed model. I had one soon after starting, but have never gone
&gt; back to look.
&gt;
&gt; All I can suggest is keep trying.
&gt;
&gt; More humor: I see you have AMD processors. The current fortran compiler is an
&gt; Intel one, which favors Intel
&gt; processors over AMD in terms of speed. And the joke is: The problem is "No
&gt; Intel Inside".
&gt;
&gt; One wonders though, if it also doesn't handle some instructions properly in
&gt; AMD processors, as a lot of users with
&gt; problems have one. Latest rumor is that the programmer has gotten hold of a
&gt; 64bit compiler.
&gt; Watch this space. Or rather, <a> href="http://www.climateprediction.net/board/viewtopic.php?t=2858"&gt; THIS</a>
&gt; space!
&gt;
&gt; Les
&gt;
&gt;
&gt;