Twice, now, I have gotten up to about 25% complete and the climate segment of BOINC freezes and dies. When it comes back, I am started from zero! What is going on??

I am on a Windows XP pro operating system, Pentium 4. Here is a copy of the last of the messages sent before it failed.

The 5/24/2005 8:51:12 PM|climateprediction.net|Sending request to scheduler: http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi
5/24/2005 8:51:13 PM|climateprediction.net|Scheduler RPC to http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi succeeded
5/24/2005 8:53:45 PM|climateprediction.net|Unrecoverable error for result 3l13_200188904_0 ( - exit code -1073741819 (0xc0000005))
5/24/2005 8:53:45 PM|climateprediction.net|Deferring communication with project for 59 seconds
5/24/2005 8:53:45 PM|climateprediction.net|Computation for result 3l13_200188904 finished
5/24/2005 8:54:45 PM|climateprediction.net|Requesting 17280.00 seconds of work
5/24/2005 8:54:45 PM|climateprediction.net|Sending request to scheduler: http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi
5/24/2005 8:54:46 PM|climateprediction.net|Scheduler RPC to http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi succeeded
5/24/2005 8:54:47 PM|climateprediction.net|Started download of 3uk5_100201375.zip
5/24/2005 8:54:48 PM|climateprediction.net|Finished download of 3uk5_100201375.zip
5/24/2005 8:54:48 PM|climateprediction.net|Throughput 20462 bytes/sec
5/24/2005 8:54:48 PM|climateprediction.net|Starting result 3uk5_100201375_0 using hadsm3 version 4.12

---

The first time was a -5 error, which is <b>usually</b> a hardware fault. The 2nd error code (-1073741819) is much more difficult to track down. A number of people are getting these at the end of one of the phases (a phase is 33% of a model).

Is your system overclocked?

Is it a Prescott or Northwood P4?

Do you have a wireless, or wired LAN, or something else?

geophi
Chuck is running multiple projects.
I wonder if it's something to do with BOINC switching projects at a bad time?
Or, at least, bad for CPDN!
Perhaps with hadsm 4.12 and certain versions of BOINC. I notice that he was also running BOINC 4.23, which I think was only ever a dev version.

Whatever the cause, 1073741819 is proliferating as bad as some 'flu strains.

Les
(Who hasn't been infected yet. Touch wood.)

geophi
Just did a google on 1073741819. 8,400 hits.
Just to cheer you up, try <a href="http://www.adminlife.com/247reference/msgs/38/191113.aspx"> this one</a> first.

Les

&gt; Just did a google on 1073741819. 8,400 hits.

Thanks for that Les. Don't claim to have read, or understood, all these, but 'access violation' and 'file handling error' were what I got out of them.

That follows. The HADSM3 app crashes in post phase processing. I've had to abandon CPDN in BOINC on my most capable machine - a 3.2 P4 running HT - because I am consistently getting this now. I had thought it might be related to the WiFi, but it's difficult to see how. Intriguingly, the Google links included reports of this same problem in Predictor. Coincidence, or are they sharing code? Or is it a BOINC related problem after all?

---

Chuck is crashing before end of phase 1, but I agree about your assessment. Also,
<a href="http://www.msusenet.com/history/topic.php/1870435911-1.html"> this one,</a> with "There is a bug, untraceable so far, ...", sounds like us.
Perhaps those with the problem can see if they have IIS, and remove / re-install it as suggested.

But if it is a program conflict between hadsm and MS, it may not be something for which there is a cure until Tolu can work it out.

Les

&gt; geophi
&gt; Chuck is running multiple projects.
&gt; I wonder if it's something to do with BOINC switching projects at a bad time?
&gt; Or, at least, bad for CPDN!
&gt;
But I've been getting this with no project switching. Athlon64 3400+ running WinXP Pro, and no other BOINC projects...no AV. I went back to my regular network card from the wifi one, and didn't get the 1073741819 error last run. But I did get a -5 error in slab, with one trickle to go in the run, the first time with a -5 in slab in 7 months.

&gt; But I've been getting this with no project switching.

Same here.

---

Narrows it down a bit. Perhaps a lot.