Short wu's?

Message boards : Number crunching : Short wu's?
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile ThePhantom86
Volunteer tester
Avatar

Send message
Joined: 30 Jan 02
Posts: 268
Credit: 1,970,082
RAC: 0
United States
Message 20682 - Posted: 1 Sep 2004, 0:35:41 UTC

I've recieved two new wu's (the first wu's I've gotten in some time) and they only take a few minutes to finish. Is there something wrong with them? Anyone else having short wu's. I wish the results table were enabled again so I could check on them. Is there any word on when that is coming back?



ID: 20682 · Report as offensive
JAF
Avatar

Send message
Joined: 9 Aug 00
Posts: 289
Credit: 168,721
RAC: 0
United States
Message 20687 - Posted: 1 Sep 2004, 1:01:46 UTC - in response to Message 20682.  

> I've recieved two new wu's (the first wu's I've gotten in some time) and they
> only take a few minutes to finish. Is there something wrong with them? Anyone
> else having short wu's. I wish the results table were enabled again so I could
> check on them. Is there any word on when that is coming back?

Not sure when all the database functions will be back. I crunched a couple WU's that were less than 3 minutes in the past two days. Probably normal.
ID: 20687 · Report as offensive
Heffed
Volunteer tester

Send message
Joined: 19 Mar 02
Posts: 1856
Credit: 40,736
RAC: 0
United States
Message 20730 - Posted: 1 Sep 2004, 2:47:37 UTC

These were most likely -9 overflows. Also known as a "noisy" WU.

After the initial scan, it is determined the resulting output file would be too large, so computation is exited. As this is not an error, you will get a few credits for these.

ID: 20730 · Report as offensive
Profile Steve Withers
Avatar

Send message
Joined: 25 Jun 99
Posts: 52
Credit: 3,083,069
RAC: 0
New Zealand
Message 20754 - Posted: 1 Sep 2004, 3:28:46 UTC - in response to Message 20682.  

> I've recieved two new wu's (the first wu's I've gotten in some time) and they
> only take a few minutes to finish. Is there something wrong with them? Anyone
> else having short wu's. I wish the results table were enabled again so I could
> check on them. Is there any word on when that is coming back?
>


>

I had a problem with short work units on the new CPDN project and found via the "Results" page that I was actually aborting the processing due to a missing glibc file on my (very stripped down) Linux mail system. The WU would abort and be reported complete - with "client error" as the status.

The CPDN results page actually included the program error reported back - including the name of the missing file! It was quickly fixed by installing the base glibc package.

Had that happened on SETI, with the results page turned off, I would never have known what the real problem was.
ID: 20754 · Report as offensive
Profile mlcudd
Volunteer tester
Avatar

Send message
Joined: 11 Apr 03
Posts: 782
Credit: 63,647
RAC: 0
United States
Message 20795 - Posted: 1 Sep 2004, 4:59:54 UTC

Hi All,
I had Full caches on all my boxes, which was fantastic after such a long drought. Unfortuantly we had a power interruption due to a storm. About 30 minutes later when the power returned and I got my boxes started back up, I had 21 WU's on one box (My Fastest AMD box), ready to report. 2 of them had times of 1.6 minutes and 2.03 minutes, however 19 of them showed no crunch time. I uploaded all that were there, they were gone in a snap, and I manually updated to remove them from my Work Tab. There were no error messages on the message tab
and as a matter of fact the only ones that showed "Download Finished" were the 2 that had miniscule times. It is the first time I have seen this. I did not get anymore WU's yet to replace those, I probably will tomorrow since that box crunches Approx 6 a day.

Everyone have a great night, and a better tomorrow.

Regards,

Rocky
ID: 20795 · Report as offensive
Profile ~misfit~

Send message
Joined: 10 Jun 99
Posts: 9
Credit: 122,638
RAC: 0
New Zealand
Message 20884 - Posted: 1 Sep 2004, 10:00:59 UTC

Since installing BOINC 4.05 I've had about one-in-five WUs processes in about half the time it normally takes. I know this isn't what this thread is initially discussing but it seemed the best place to post. I've had 'noisy' units before, rarely, where processing is aborted after a few minutes but this is different. About 20% of my WUs are showing 100% processed in around 50% of the time expected since installing BOINC4.05/SETI 4.03. This is over about 20 WUs processed since installing 4.05 and seems consistant.
A bug perhaps?
Win XP SP1. Athlon XP on an nForce2 board, 512MB RAM.
ID: 20884 · Report as offensive
Profile mlcudd
Volunteer tester
Avatar

Send message
Joined: 11 Apr 03
Posts: 782
Credit: 63,647
RAC: 0
United States
Message 20905 - Posted: 1 Sep 2004, 12:28:48 UTC

Hi All,
I got some new WU's last night to cover the ones actually crunched and the ones that crunched in lighting speed. However the same exact thing just happened again, only fewer in number. I had one unit to "Transfer", that uploaded as should, but on the Work Tab, there wer 6 WU's Ready to report, only one of them with "Actual Real Time." The rest had "0" time.
Maybe the fewer number indicates they are being flushed out of the system (I hope).

Happy Crunching!!

Regards,

Rocky
ID: 20905 · Report as offensive
Profile ~misfit~

Send message
Joined: 10 Jun 99
Posts: 9
Credit: 122,638
RAC: 0
New Zealand
Message 22038 - Posted: 3 Sep 2004, 22:17:54 UTC

Since my last post in this thread I've been getting a situation where about 60% of my WUs show either just under or just over *one minute*. The usual completeion time is 2:45. I have this situation on at least two machines and on the one worst affected I left Prime95 running overnight to test the stability of my machine. Prime ran for eight hours 35 minutes with no errors. So the machine's OK, I guess that leaves BOINC? I can't imagine, going by the percentage of 'noisy units' with classic, that these are all noisy units. With classic I'd see maybe one in 60 WUs being noisy, not two out of three.
ID: 22038 · Report as offensive
Profile The worm that turned
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 100
Credit: 4,872,533
RAC: 0
Australia
Message 22044 - Posted: 3 Sep 2004, 22:26:37 UTC
Last modified: 10 Sep 2004, 2:02:00 UTC

Since installing 4.05 i've also had a far higher % of fast units than before.
Perhaps as high as 10% .
No error messages with them so I suppose they are just 'noisy' units.

ID: 22044 · Report as offensive
Profile ~misfit~

Send message
Joined: 10 Jun 99
Posts: 9
Credit: 122,638
RAC: 0
New Zealand
Message 22075 - Posted: 3 Sep 2004, 23:04:38 UTC - in response to Message 22038.  

> Since my last post in this thread I've been getting a situation where about
> 60% of my WUs show either just under or just over *one minute*. The usual
> completeion time is 2:45. I have this situation on at least two machines and
> on the one worst affected I left Prime95 running overnight to test the
> stability of my machine. Prime ran for eight hours 35 minutes with no errors.
> So the machine's OK, I guess that leaves BOINC? I can't imagine, going by the
> percentage of 'noisy units' with classic, that these are all noisy units. With
> classic I'd see maybe one in 60 WUs being noisy, not two out of three.

On further investigation it seems that the above-mentioned WUs were all part of a nearly-contigous block of units with numbers in the range of 04my04aa.25870.22832.872154.2xx_x, I had 20 of them, of which only four completed as expected. The rest all show between 58 secs and 1 min 29 secs. Since that block of WUs has been 'completed' it seems all is back to normal. Had me worried there.
ID: 22075 · Report as offensive
Profile Sir Ulli
Volunteer tester
Avatar

Send message
Joined: 21 Oct 99
Posts: 2246
Credit: 6,136,250
RAC: 0
Germany
Message 22081 - Posted: 3 Sep 2004, 23:13:49 UTC
Last modified: 3 Sep 2004, 23:17:19 UTC

on my AMD64 3.200+ i have at 10 Units 2 that are fast, ...

on Seti@home classic wi call this

Sweet Wus ... :)

Greetings from Germany NRW
Ulli [/url]

ID: 22081 · Report as offensive
[boinc.at] jbs
Avatar

Send message
Joined: 17 Dec 99
Posts: 40
Credit: 1,001,969
RAC: 0
Austria
Message 22415 - Posted: 4 Sep 2004, 21:58:38 UTC - in response to Message 22075.  

> On further investigation it seems that the above-mentioned WUs were all part
> of a nearly-contigous block of units with numbers in the range of
> 04my04aa.25870.22832.872154.2xx_x, I had 20 of them, of which only four
> completed as expected. The rest all show between 58 secs and 1 min 29 secs.
> Since that block of WUs has been 'completed' it seems all is back to normal.
> Had me worried there.
>
>

And 25ap04aa.....
ID: 22415 · Report as offensive
Profile Thierry Van Driessche
Volunteer tester
Avatar

Send message
Joined: 20 Aug 02
Posts: 3083
Credit: 150,096
RAC: 0
Belgium
Message 22422 - Posted: 4 Sep 2004, 22:14:07 UTC
Last modified: 4 Sep 2004, 22:15:10 UTC

I have normal CPU time between 3:45h. and 4:35h using HT.

New records broken here:
WU 25ap04aa.23553.18161.92308.xxx had a CPU time of 6:53:27h.
WU 04my04aa.2277.12866.972134.xxx had a CPU time of 00:00:00h.

Greetings from Belgium
ID: 22422 · Report as offensive
Heffed
Volunteer tester

Send message
Joined: 19 Mar 02
Posts: 1856
Credit: 40,736
RAC: 0
United States
Message 22481 - Posted: 5 Sep 2004, 2:54:44 UTC - in response to Message 22038.  

> Since my last post in this thread I've been getting a situation where about
> 60% of my WUs show either just under or just over *one minute*. The usual
> completeion time is 2:45. I have this situation on at least two machines and
> on the one worst affected I left Prime95 running overnight to test the
> stability of my machine. Prime ran for eight hours 35 minutes with no errors.
> So the machine's OK, I guess that leaves BOINC? I can't imagine, going by the
> percentage of 'noisy units' with classic, that these are all noisy units. With
> classic I'd see maybe one in 60 WUs being noisy, not two out of three.

What does your stderr file show completion errors?

ID: 22481 · Report as offensive
texasfit
Avatar

Send message
Joined: 11 May 03
Posts: 223
Credit: 500,626
RAC: 0
United States
Message 22492 - Posted: 5 Sep 2004, 3:34:39 UTC

>>>> What does your stderr file show completion errors?

========================================================

I also have some of these wu's with completion times of 1-3 minutes.
Currently there are 56 wu's 'Ready to Report' and 11 are showing the
quick completion time. There are no errors showing in the stderr.txt
file. In fact my stderr file does not have any entries at all.

ID: 22492 · Report as offensive
Profile Carl Cuseo
Volunteer tester
Avatar

Send message
Joined: 18 Jan 02
Posts: 652
Credit: 34,312
RAC: 0
Puerto Rico
Message 22497 - Posted: 5 Sep 2004, 4:02:20 UTC - in response to Message 22492.  

About 20% of WU's I have ready to report are short-timers.
ALL are from 25 April...cc
ID: 22497 · Report as offensive
Profile ~misfit~

Send message
Joined: 10 Jun 99
Posts: 9
Credit: 122,638
RAC: 0
New Zealand
Message 22601 - Posted: 5 Sep 2004, 13:17:39 UTC - in response to Message 22481.  

>
> What does your stderr file show completion errors?

Where do I find it and how do I read it?

misfit(at)orcon(dot)net(dot)nz in case I can't get back here, there have been outages, at least from my end.
ID: 22601 · Report as offensive
Profile ~misfit~

Send message
Joined: 10 Jun 99
Posts: 9
Credit: 122,638
RAC: 0
New Zealand
Message 22603 - Posted: 5 Sep 2004, 13:26:24 UTC - in response to Message 22601.  

> >
> > What does your stderr file show completion errors?
>
> Where do I find it and how do I read it?
>
> misfit(at)orcon(dot)net(dot)nz in case I can't get back here, there have been
> outages, at least from my end.

Found it. It doesn't show any errors, it only shows the same messages that are in the client tab under 'messages' and that is automatically trimmed somehow so the messages from when those WUs finished are gone. My last ten WUs have taken the 'normal' time (2:40:00 or thereabouts) and the stderr just shows a mass of errors from the software trying to connect to the servers, basically this:

2004-09-05 10:20:37 [SETI@home] Scheduler RPC to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi failed
2004-09-05 10:20:37 [SETI@home] No schedulers responded
2004-09-05 10:20:37 [SETI@home] Deferring communication with project for 1 minutes and 0 seconds

Repeated ad-infinitum.
ID: 22603 · Report as offensive
Profile ~misfit~

Send message
Joined: 10 Jun 99
Posts: 9
Credit: 122,638
RAC: 0
New Zealand
Message 22605 - Posted: 5 Sep 2004, 13:30:32 UTC - in response to Message 22081.  

> on my AMD64 3.200+ i have at 10 Units 2 that are fast, ...
>
> on Seti@home classic wi call this
>
> Sweet Wus ... :)

However, on SETI classic it was one credit for one WU, BOINC is different as you know. On SETI-BOINC we call these fast WUs....

A pain in the arse. :-)

As they mess up cache size/time-to-run and I don't think we'll get any credit for them.
ID: 22605 · Report as offensive

Message boards : Number crunching : Short wu's?


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.