climateprediction.net home page
Client errors. ( I did search too....:))

Client errors. ( I did search too....:))

Questions and Answers : Windows : Client errors. ( I did search too....:))
Message board moderation

To post messages, you must log in.

AuthorMessage
old_user65799

Send message
Joined: 21 Mar 05
Posts: 13
Credit: 17,566
RAC: 0
Message 11240 - Posted: 21 Mar 2005, 15:45:58 UTC
Last modified: 21 Mar 2005, 15:48:08 UTC

Hey guys/gals, I just decided to try this Climate prediction here to keep CPU busy cause Seti is down.
So I started on my XP2100+ machine and lost 2 WU's right off the bat. I get this error, 4.25
- exit code -5 (0xfffffffb)

I been searching all morning and havent really found a good answer. (and I suck at searching)
Yes, the PC is overclocked, and yes I have tested with Prime95 and Memtest86 for 48 hours each, manditory once a month to make sure all is good.
I was running Folding@Home, for quite some time, but decided to go back to do some Seti.
Machine details are:
ABIT NF7-s v.1 (200mhz bus)
1gig Mushkin 2.5-3-3-11
AMD XP2100+@2200mhz
Albatron GeForce 4200Ti (not o/c)
If this machine is stable in EVERYTHING else
The CPU never gets above 35°C, it is water cooled.
I have heatsinks on the 'fets, extra fans blowing in the case, I am a cooling overkiller so I really dont think she's overheating.
And this project looks really cool, and I would really like to run it.
Any ideas what might be going on here????

EDIT:
I should also mention, I stopped running it (disconnected from the project) simply because I didnt want to waste your WU's) unit I can figure something out...
ID: 11240 · Report as offensive     Reply Quote
Profile geophi
Volunteer moderator

Send message
Joined: 7 Aug 04
Posts: 2167
Credit: 64,403,322
RAC: 5,085
Message 11241 - Posted: 21 Mar 2005, 15:58:34 UTC

This is something I'm sure you won't be good with long term, but just to test, I'd downclock it to under 2 GHz and see if it will run stable. Normally if you can run Prime95 and Memtest86+ stably for <b>long</b> periods of time, you should be stable in CPDN, but I don't know what else to tell you. -5 is a computation error where some grid point temperature in the model goes to absolute zero, usually the result of a machine error, although possibly a bad set of parameters. But if you get this repeatedly, it could not be the parameters.
ID: 11241 · Report as offensive     Reply Quote
old_user65799

Send message
Joined: 21 Mar 05
Posts: 13
Credit: 17,566
RAC: 0
Message 11248 - Posted: 21 Mar 2005, 18:02:41 UTC - in response to Message 11241.  

&gt; This is something I'm sure you won't be good with long term, but just to test,
&gt; I'd downclock it to under 2 GHz and see if it will run stable. Normally if
&gt; you can run Prime95 and Memtest86+ stably for <b>long</b> periods of time, you
&gt; should be stable in CPDN, but I don't know what else to tell you. -5 is a
&gt; computation error where some grid point temperature in the model goes to
&gt; absolute zero, usually the result of a machine error, although possibly a bad
&gt; set of parameters. But if you get this repeatedly, it could not be the
&gt; parameters.
&gt;

Tryed lowering it, same results.
Turned the bus down to 166 same, 133 same.
A few times I got the window about some had3m or what ever it was, error'd out and ask'd to send to M$, at which point I suspended the work, told it not to send, restarted BOINC and resumed the WU, which it did, then would error out again, with the hadthingy error, finally the last time it didnt do that, it just said the error and sending results.
I'll run memtest and prime again to see what happins.
I would try it on one of my other machines, but I dont want to waste work units......:D
Perhaps I am not ment to run this......:(
ID: 11248 · Report as offensive     Reply Quote
old_user65799

Send message
Joined: 21 Mar 05
Posts: 13
Credit: 17,566
RAC: 0
Message 11254 - Posted: 21 Mar 2005, 20:33:00 UTC
Last modified: 21 Mar 2005, 20:34:22 UTC

Set everything in bios the pertains to CPU and voltage to default :( and it's working now.
I will have to get to the bottom of that in a big hurry.....

Thanks for your help, I forgot to mention that before.......
ID: 11254 · Report as offensive     Reply Quote
old_user65799

Send message
Joined: 21 Mar 05
Posts: 13
Credit: 17,566
RAC: 0
Message 11275 - Posted: 22 Mar 2005, 11:56:18 UTC

It worked for a while......
Something isnt right here.....
ID: 11275 · Report as offensive     Reply Quote
Profile old_user28658
Avatar

Send message
Joined: 5 Nov 04
Posts: 17
Credit: 85,526
RAC: 0
Message 11280 - Posted: 22 Mar 2005, 14:51:05 UTC

Unfortunately, as you've realised, CPDN is the most punishing and demanding of the BOINC projects.

It will find errors that most diagnostic software doesn't even look for :(

good luck and keep BIONCing away :)
ID: 11280 · Report as offensive     Reply Quote
old_user65799

Send message
Joined: 21 Mar 05
Posts: 13
Credit: 17,566
RAC: 0
Message 11289 - Posted: 22 Mar 2005, 21:34:06 UTC



Gonna give it one more try. Hopefully the things I have done will help....
If it craps out, I am not gonna waste the time of the developers with incomplete work units.
ID: 11289 · Report as offensive     Reply Quote
old_user23880
Volunteer tester

Send message
Joined: 10 Oct 04
Posts: 223
Credit: 4,664
RAC: 0
Message 11298 - Posted: 23 Mar 2005, 3:35:33 UTC

The incomplete WUs go back into the pool and are reissued sooner or later, so they're not lost.

Make sure that when you exit cpdn, you suspend the model first and give it plenty of time to close down.

A solution of last resort which I have adopted because of repeated -5 errors is to run classic cpdn instead of boinc cpdn. Classic almost always runs stably and at about the same speed as boinc, you see the visualisations, can post on all the boards, contribute useful models and learn a bit about climate as you go. The disadvantages are that you can't share computer time with other boinc projects and your classic stats don't get added to your boinc total.

If you want to try this even for just 1 model, get to it from the cpdn home page, click on Open Uni course, and you'll see the download button for classic.

I uninstalled boinc first to avoid having 2 cpdn systems (classic + boinc)in my files simultaneously, and also because when I give boinc another try sometime in the future, I will then get the newest version. I don't actually know, however, whether it really is necessary to uninstall boinc before installing classic, so maybe someone with more knowledge could comment on this.
__________________________________________________

ID: 11298 · Report as offensive     Reply Quote
old_user65799

Send message
Joined: 21 Mar 05
Posts: 13
Credit: 17,566
RAC: 0
Message 11331 - Posted: 23 Mar 2005, 22:06:35 UTC

Must have done something right, I got a trickle......


(that sounds funny)
ID: 11331 · Report as offensive     Reply Quote
Helmer Bryd

Send message
Joined: 16 Aug 04
Posts: 147
Credit: 7,748,561
RAC: 8,366
Message 11338 - Posted: 23 Mar 2005, 22:54:23 UTC

I don't think the default voltage is enough.
Try 2.9v on the DIMMs and up the vcore some.

(worked for me anyway on my Abit NF 7)
ID: 11338 · Report as offensive     Reply Quote
Profile Friedrich S.
Avatar

Send message
Joined: 22 Jan 05
Posts: 35
Credit: 3,826,027
RAC: 2,069
Message 11368 - Posted: 24 Mar 2005, 16:50:04 UTC

Hi,

I had the first three models crash, too. All with this famous exit code -5, mostly well below 15% or the work done. My system definitely is stable. Never ever any stability related problem with any application.

Then somebody suggested - and I tried - to regularely defragment the drive. (Of course while BOINC is not running!)
Since then the current two models have been running fine, currently at slightly above and below 35%. I keep defragmenting it about every two weeks, and everything seems fine. (Keep my fingers crossed!)

I am not sure what that means, what exactly was the problems and why it should make a difference. But since it is a real easy thing to do, it might be worth a try.

Friedrich



I love CPDN, but I have only Windows clients.
--
ID: 11368 · Report as offensive     Reply Quote
old_user65799

Send message
Joined: 21 Mar 05
Posts: 13
Credit: 17,566
RAC: 0
Message 11420 - Posted: 25 Mar 2005, 19:18:03 UTC

I use Diskeeper and have it scheduled to go once a week.
More often on frequently used PC's (my main PC does daily).
ID: 11420 · Report as offensive     Reply Quote

Questions and Answers : Windows : Client errors. ( I did search too....:))

©2024 climateprediction.net