Boinc version 4.43.

Firstly Boinc decided to download 2 seperate WU one day appart (given that CPDN units take about 6+ months to run when sharing resources with other projects, the 2nd one would likely never finish in time). This was about a week ago just after I upgraded from 4.17 to 4.43. One downloaded before the update, one after.

This sets the stage for the even stranger error, Boinc uploading BOTH workunits before completion. I hesitate to do any further WUs if this is going to keep happening.

WU 1 (3rim_200197394_0) uploaded with about 4% completion.
WU 2 (439c_100212762_0) uploaded with no work done (0 seconds).



6/8/2005 5:13:17 PM|climateprediction.net|Finished upload of 3rim_200197394_0_1.zip
6/8/2005 5:13:17 PM|climateprediction.net|Throughput 121 bytes/sec
6/8/2005 5:13:17 PM|climateprediction.net|Started upload of 3rim_200197394_0_3.zip
6/8/2005 5:13:23 PM|climateprediction.net|Finished upload of 3rim_200197394_0_2.zip
6/8/2005 5:13:23 PM|climateprediction.net|Throughput 3405 bytes/sec
6/8/2005 5:13:23 PM|climateprediction.net|Started upload of 3rim_200197394_0_4.zip
6/8/2005 5:13:38 PM|climateprediction.net|Finished upload of 3rim_200197394_0_3.zip
6/8/2005 5:13:38 PM|climateprediction.net|Throughput 167 bytes/sec
6/8/2005 5:13:38 PM|climateprediction.net|Started upload of 3rim_200197394_0_5.zip
6/8/2005 5:13:41 PM|climateprediction.net|Finished upload of 3rim_200197394_0_4.zip
6/8/2005 5:13:41 PM|climateprediction.net|Throughput 223 bytes/sec
6/8/2005 5:13:42 PM|climateprediction.net|Started upload of 439c_100212762_0_1.zip
6/8/2005 5:13:46 PM|climateprediction.net|Finished upload of 3rim_200197394_0_5.zip
6/8/2005 5:13:46 PM|climateprediction.net|Throughput 586 bytes/sec
6/8/2005 5:13:46 PM|climateprediction.net|Started upload of 439c_100212762_0_2.zip
6/8/2005 5:13:47 PM|climateprediction.net|Finished upload of 439c_100212762_0_1.zip
6/8/2005 5:13:47 PM|climateprediction.net|Throughput 0 bytes/sec
6/8/2005 5:13:47 PM|climateprediction.net|Started upload of 439c_100212762_0_3.zip
6/8/2005 5:13:48 PM|climateprediction.net|Finished upload of 439c_100212762_0_2.zip
6/8/2005 5:13:48 PM|climateprediction.net|Throughput 0 bytes/sec
6/8/2005 5:13:48 PM|climateprediction.net|Finished upload of 439c_100212762_0_3.zip
6/8/2005 5:13:48 PM|climateprediction.net|Throughput 0 bytes/sec
6/8/2005 5:13:48 PM|climateprediction.net|Started upload of 439c_100212762_0_4.zip
6/8/2005 5:13:48 PM|climateprediction.net|Started upload of 439c_100212762_0_5.zip
6/8/2005 5:13:52 PM|climateprediction.net|Finished upload of 439c_100212762_0_4.zip
6/8/2005 5:13:52 PM|climateprediction.net|Throughput 0 bytes/sec
6/8/2005 5:13:53 PM|climateprediction.net|Finished upload of 439c_100212762_0_5.zip
6/8/2005 5:13:53 PM|climateprediction.net|Throughput 0 bytes/sec


Checking the WUs from my Results stream:

For the 3rim one:
<core_client_version>4.43</core_client_version>
<message> - exit code -5 (0xfffffffb)
</message>

Other one isn't listed at all.


--RC

The second download is a bug in 4.43 which is believed to be fixed in 4.44. It sounds like the 439c unit was misallocated so it was good to get rid of it.

Unfortunately -5 is a general catch-all for computation error. Some AMD computers seem to have problems despite lots of stability testing. Sometimes it is down to hardware problems but I see you have completed a full model. If the weather has warmed up where you are, you might consider some maintenance
http://www.climateprediction.net/board/viewtopic.php?t=2124 and/or stability checks http://www.climateprediction.net/board/viewtopic.php?t=2126

Hope it turns out to be a one off occurence for you.

The extra download is a bug in BOINC 4.43, which is fixed in 4.44

3rim_200197394_0_1 finished early because it failed, with error code -5, a code given to misc failures not covered by other codes.

The 2nd model is, indeed, mysterious.

Some things to check: DON"T overclock; if you have an ATI graphics card, some of these are known to cause problems; hardware checks <a href="http://www.climateprediction.net/board/viewtopic.php?t=2126"> here</a> and <a href="http://www.climateprediction.net/board/viewtopic.php?t=2124"> here;</a> malware checkers <a href="http://www.climateprediction.net/board/viewtopic.php?t=2895"> here.</a>

What is your CP side of things doing now?

Les

Thanks for the help, I figured that the first one might have been a computation error, but the 2nd one was what really threw me. Perhaps when it uploaded the 1st CPDN said it shouldn't have the 2nd?

-------
To answer Les's question:
CP right now is on "don't get more work" (I set that in Boinc). I'll upgrade to 4.44 before telling it to get more WUs (EDIT: have to wait for it to be released still it seems—not on the BOINC download site yet).

-----
To give more info for both of your answers:
No overclocking. Fans and heatsink cleaned only about 2 weeks ago. No malware (I keep a very tight control on what runs on my system).

My rig is an AMD Athlon XP 3000+ on an Asus A7N8X-Deluxe mobo, nVidia gForce 4 TI 4200 graphics card, 1GB ram (PC3200 DDR in dual channel, 0.5GB/channel), 465W Power Supply, and more fans than any sane person needs :). I do have an ATI TV Tuner on it though. As for AV, I'm running Norton Corporate Edition (Prog ver: 7.60.926, Scan engine ver: 4.1.0.15).

CPU maxes at about 48°C when under full load (occasionally it goes over that...and I think it did this past weekend when the ambient temp in the house hit about 82°F becuase I didn't have the AC on, but I know the CPU didn't break 55°C becuase I'd have gotten a warning if it had). Given that the die supports up to 85°C (that's from memory, might be off), there shouldn't be any stability issues there.

Just ran the voltage monitor on it too. Stable voltage between 100% CPU load and about 25% load. 12V and 5V lines are all very close to perfect. 3.3V is a little high and almost out of bounds, but it is stable and not breaking the upper limit (assuming 5% tolerence). VCore is stable (and I believe close to perfect from memory, but to confirm that would require looking up the data on what it should be).

Given that the first WU that actually downloaded (if you look at my "Results" I have about 5 WUs that had download problems back when CPDN first got their Boinc version going) successfully ran all the way to the end...I kind of doubt that the rig itself has any stability problem under CPDN's app.

One thing I was doing when it crashed was lots of renders in Blender, and I didn't always remember to turn Boinc to "suspend" before rendering. It is possible that had something to do with it since when rendering there is very high CPU and memory usage from Blender.

BTW, on your system tests post they mention SuperPi. I've run that in the past (about a year ago) with no problems to 33 million digits.

&gt; Some things to check: DON"T overclock; if you have an ATI graphics card, some
&gt; of these are known to cause problems; hardware checks <a> href="http://www.climateprediction.net/board/viewtopic.php?t=2126"&gt; here</a>
&gt; and <a href="http://www.climateprediction.net/board/viewtopic.php?t=2124">
&gt; here;</a> malware checkers <a> href="http://www.climateprediction.net/board/viewtopic.php?t=2895"&gt; here.</a>
&gt;
&gt; What is your CP side of things doing now?
&gt;
&gt; Les

&gt; To answer Les's question:
&gt; CP right now is on "don't get more work" (I set that in Boinc). I'll upgrade
&gt; to 4.44 before telling it to get more WUs (EDIT: have to wait for it to be
&gt; released still it seems—not on the BOINC download site yet).

Hmmm. It was there last week, but I do know that the version number has gone up to at least 4.45. You could try copying the link and changing the '43'.
<br><a href="http://www.teampicard.net/"><img src="http://www.teampicard.net/images/picardmini.gif"></a><a href="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/team_display.php?teamid=3">Join us here</a>

Strange. It was there with changing the link manually. Boinc.berkley.edu just isn't updating their download page...

&gt; Hmmm. It was there last week, but I do know that the version number has gone
&gt; up to at least 4.45. You could try copying the link and changing the '43'.
&gt; <br><a href="http://www.teampicard.net/"><img> src="http://www.teampicard.net/images/picardmini.gif"&gt;</a><a> href="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/team_display.php?teamid=3"&gt;Join
&gt; us here</a>
&gt;

&gt;Boinc.berkley.edu just isn't updating their download page...

They sometimes drop back on that page to an earlier version, presumably if the latest is reported to have serious bugs, or perhaps if another dev version is about to emerge.

Ok, I updated to 4.45 and turned downloads of CPDN back on.

And it isn't requesting any data?

6/9/2005 11:34:50 AM|climateprediction.net|Requesting 0 seconds of work, returning 0 results

I'm confused :) - is "Boinc" screwing up again?

Maybe I should have stayed with 4.17. It worked better and didn't have the behemouth of a system hog that is "Boinc Manager."

There are reasons why CPDN still offers 4.19 as the only recommended version :)

That isn't an option if you are running some other projects (notably Einstein I believe).

&gt; There are reasons why CPDN still offers 4.19 as the only recommended version
&gt; :)
&gt;
&gt; That isn't an option if you are running some other projects (notably Einstein
&gt; I believe).
&gt;
&gt;Andrew: That is an interesting comment. I re-installed BOINC 4.19 in April and have not seen any of the disturbances which occurred under 4.25. Oh, except for a lost trickle, (Phase 1 of all places) which does not YET seem to have affected the completion of the WU. Les Bayliss is helping me thru this crisis.

Q: In the case of CPDN, is there some desperately compelling reason to upgrade, or is the compulsion like Everest, it is just there?