Advanced search

Forums : Technical Support : Invalid results
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · 4 · Next

AuthorMessage
Profile [AF>Futura Sciences>Linux] Thrr-Gilag
Volunteer tester
Avatar

Send message
Joined: 22 May 07
Posts: 32
Credit: 145,576
RAC: 0
Message 279 - Posted: 19 Jun 2007, 22:51:22 UTC

I have noticed that my Pentium 2 host (http://cosmos.astro.uiuc.edu/cosmohome/results.php?hostid=66) has a lot of invalid results in regard of my other hosts.

The computation ends each time good for BOINC client. There is no other work on this host that could explain a switch between different WU.

So is this a bug for with validation, a problem of accurency results or the pentium 2 architecture is below the requirement (but it gave good results before) ?
------
Thrr-Gilag Kee'rr

L'Alliance Francophone
ID: 279 · Report as offensive     Reply Quote
Profile Scott
Volunteer moderator
Project administrator
Project developer
Avatar

Send message
Joined: 1 Apr 07
Posts: 662
Credit: 13,742
RAC: 0
Message 287 - Posted: 20 Jun 2007, 14:35:40 UTC - in response to Message 279.  
Last modified: 20 Jun 2007, 15:01:58 UTC

I have noticed that my Pentium 2 host (http://cosmos.astro.uiuc.edu/cosmohome/results.php?hostid=66) has a lot of invalid results in regard of my other hosts.

The computation ends each time good for BOINC client. There is no other work on this host that could explain a switch between different WU.

So is this a bug for with validation, a problem of accurency results or the pentium 2 architecture is below the requirement (but it gave good results before) ?

The problem is due to the validator. The issue is that results have to agree very closely (to 1 part in 10^4, actually). Since different CPUs do floating-point arithmetic differently, some results will not agree.

I am trying to push to a) relax the accuracy requirements and/or b) use homogeneous redundancy. Right now, we can't use HR because we have few enough machines that some work might not ever be done. Once we initiate the next testing phase (which will feature open registration), we can start using it.

However, in the present, we're going to have to deal with the reality that a good portion of WUs will be invalidated while we're testing our validator application. Trust me, though, that once HR is turned on, the percentage of invalid results should drop below 1%.

EDIT: I made a change to the validator that increases the maximum number of "invalid" rows allowed from 1 to 20. This should greatly reduce the frequency of invalidated results.
Scott Kruger
Project Administrator, Cosmology@Home
ID: 287 · Report as offensive     Reply Quote
Profile [AF>Futura Sciences>Linux] Thrr-Gilag
Volunteer tester
Avatar

Send message
Joined: 22 May 07
Posts: 32
Credit: 145,576
RAC: 0
Message 289 - Posted: 20 Jun 2007, 17:28:17 UTC

thancks a lot for your explanation ^^
------
Thrr-Gilag Kee'rr

L'Alliance Francophone
ID: 289 · Report as offensive     Reply Quote
Modesto

Send message
Joined: 22 Jun 07
Posts: 2
Credit: 7,794
RAC: 0
Message 334 - Posted: 22 Jun 2007, 23:38:21 UTC

This is just feedback not a complaint...

I had one invalid result on my only host, which ended successfully according to the Boinc client. The other 71 results listed are either ok or still pending.
ID: 334 · Report as offensive     Reply Quote
Profile Saenger
Volunteer tester
Avatar

Send message
Joined: 22 May 07
Posts: 110
Credit: 282,157
RAC: 0
Message 335 - Posted: 23 Jun 2007, 8:12:24 UTC

Same as Modesto, only for information, no complaint (I choose to test this, so I wanted to loose credit ;)

I've got 6 invalids so far from the last batch for my puter, I know there were some from the first batch as well, but the deleter seems to have done his work yet.

So I'm waiting for HR (and open account creation) to get more credits. But as long as your WUs don't shred my puter I can easily live with the (not quite) invalids.
Grüße vom Sänger
ID: 335 · Report as offensive     Reply Quote
Profile [B^S] Acmefrog
Volunteer tester
Avatar

Send message
Joined: 8 Jun 07
Posts: 175
Credit: 446,074
RAC: 0
Message 342 - Posted: 23 Jun 2007, 18:41:17 UTC

I just had 4 invalids in a row on one fo my pc's. It also has a few pending. I have suspended it until I can see if those will be invalid. It might just be a case of bad luck because other WUs have been fine.
ID: 342 · Report as offensive     Reply Quote
Stwainer

Send message
Joined: 21 Jun 07
Posts: 18
Credit: 536,245
RAC: 0
Message 346 - Posted: 23 Jun 2007, 23:50:56 UTC

Seems like my invalid results run in packs as well. I get a bunch of valid ones, then a patch of those that are invalid. Is there a trend or is it just me reading more into the little data I have?
ID: 346 · Report as offensive     Reply Quote
Profile KSMarksPsych
Volunteer tester
Avatar

Send message
Joined: 22 May 07
Posts: 92
Credit: 57,682
RAC: 0
Message 348 - Posted: 24 Jun 2007, 2:29:04 UTC

It looks like I'm getting quite a few on my linux host that are either pending with multiple new copies going our or invalid.

I've been running Cosmology exclusively on it for about 18 hours or so now.

I'm just trying to clear out the backlog I have from when I changed my local preferences on the machine.

*eek* sorry about that.

I'm churning through them at about 5 minutes a piece, two at a time.
Kathryn :o)
The BOINC FAQ Service
The Unofficial BOINC Wiki
The Trac System
More BOINC information than you can shake a stick of RAM at.
ID: 348 · Report as offensive     Reply Quote
Profile [B^S] Acmefrog
Volunteer tester
Avatar

Send message
Joined: 8 Jun 07
Posts: 175
Credit: 446,074
RAC: 0
Message 351 - Posted: 24 Jun 2007, 4:26:39 UTC

I went ahead and decided to let my pc crunch more WUs to see if the invalid results kept happening. Well just my luck, of the 12 or so results I have returned, they are all still pending.

Also it seems that these WUs are sent to 4 people initially. I wonder why this one had an initial replication of 6.
http://cosmos.astro.uiuc.edu/cosmohome/workunit.php?wuid=86420
ID: 351 · Report as offensive     Reply Quote
Profile [B^S] Acmefrog
Volunteer tester
Avatar

Send message
Joined: 8 Jun 07
Posts: 175
Credit: 446,074
RAC: 0
Message 352 - Posted: 24 Jun 2007, 5:38:10 UTC
Last modified: 24 Jun 2007, 5:43:13 UTC

Odd this one has 8
http://cosmos.astro.uiuc.edu/cosmohome/workunit.php?wuid=86856

Now it created another WU and the initial now says 9.
ID: 352 · Report as offensive     Reply Quote
Profile KSMarksPsych
Volunteer tester
Avatar

Send message
Joined: 22 May 07
Posts: 92
Credit: 57,682
RAC: 0
Message 353 - Posted: 24 Jun 2007, 6:08:35 UTC - in response to Message 352.  

Odd this one has 8
http://cosmos.astro.uiuc.edu/cosmohome/workunit.php?wuid=86856

Now it created another WU and the initial now says 9.



ouch

That's a new one for me.

I've never seen the initial replication field changing once the WU was created.

A script gone amuck?
Kathryn :o)
The BOINC FAQ Service
The Unofficial BOINC Wiki
The Trac System
More BOINC information than you can shake a stick of RAM at.
ID: 353 · Report as offensive     Reply Quote
Profile vfrey

Send message
Joined: 24 Jun 07
Posts: 4
Credit: 39,046
RAC: 4
Message 358 - Posted: 24 Jun 2007, 13:45:35 UTC

this might be the reason for this:

http://cosmos.astro.uiuc.edu/cosmohome/workunit.php?wuid=86936
ID: 358 · Report as offensive     Reply Quote
Profile [B^S] Acmefrog
Volunteer tester
Avatar

Send message
Joined: 8 Jun 07
Posts: 175
Credit: 446,074
RAC: 0
Message 359 - Posted: 24 Jun 2007, 17:51:54 UTC

ID: 359 · Report as offensive     Reply Quote
Profile [B^S] Acmefrog
Volunteer tester
Avatar

Send message
Joined: 8 Jun 07
Posts: 175
Credit: 446,074
RAC: 0
Message 360 - Posted: 24 Jun 2007, 17:53:08 UTC - in response to Message 351.  

I went ahead and decided to let my pc crunch more WUs to see if the invalid results kept happening. Well just my luck, of the 12 or so results I have returned, they are all still pending.

Also it seems that these WUs are sent to 4 people initially. I wonder why this one had an initial replication of 6.
http://cosmos.astro.uiuc.edu/cosmohome/workunit.php?wuid=86420


This one ended with too many successes as well
ID: 360 · Report as offensive     Reply Quote
Marky-UK
Volunteer tester

Send message
Joined: 8 Jun 07
Posts: 9
Credit: 18,440
RAC: 0
Message 362 - Posted: 24 Jun 2007, 19:02:02 UTC - in response to Message 359.  
Last modified: 24 Jun 2007, 19:02:46 UTC

Well some of these had too many success results (not sure what that means).
http://cosmos.astro.uiuc.edu/cosmohome/workunit.php?wuid=87385
and
http://cosmos.astro.uiuc.edu/cosmohome/workunit.php?wuid=86180

Something is very wrong with the server (validator?) - it just seems to keep sending the same WUs over and over (even though they succeed). Eventually the 'maximum number of success results' figure (10) is reached and the whole lot gets 0 credit.
ID: 362 · Report as offensive     Reply Quote
Profile Misfit
Volunteer tester
Avatar

Send message
Joined: 9 Jun 07
Posts: 150
Credit: 237,789
RAC: 0
Message 363 - Posted: 24 Jun 2007, 20:15:32 UTC
Last modified: 24 Jun 2007, 20:18:28 UTC

http://cosmos.astro.uiuc.edu/cosmohome/workunit.php?wuid=87346
http://cosmos.astro.uiuc.edu/cosmohome/workunit.php?wuid=86180
errors Too many success results

With many more WUs successful, quorum met, still pending.
Pending credit: 119.26
me@rescam.org
ID: 363 · Report as offensive     Reply Quote
Profile [B^S] Acmefrog
Volunteer tester
Avatar

Send message
Joined: 8 Jun 07
Posts: 175
Credit: 446,074
RAC: 0
Message 365 - Posted: 24 Jun 2007, 20:38:43 UTC - in response to Message 362.  

Well some of these had too many success results (not sure what that means).
http://cosmos.astro.uiuc.edu/cosmohome/workunit.php?wuid=87385
and
http://cosmos.astro.uiuc.edu/cosmohome/workunit.php?wuid=86180

Something is very wrong with the server (validator?) - it just seems to keep sending the same WUs over and over (even though they succeed). Eventually the 'maximum number of success results' figure (10) is reached and the whole lot gets 0 credit.


I agree something isn't working properly. I'm not savvy enough to know if it is a batch of bad WUs or something with the Cosmo servers or what. I've seen many more WUs with an increasing initial run numbers.

ID: 365 · Report as offensive     Reply Quote
Modesto

Send message
Joined: 22 Jun 07
Posts: 2
Credit: 7,794
RAC: 0
Message 373 - Posted: 25 Jun 2007, 2:53:09 UTC - in response to Message 362.  

Well some of these had too many success results (not sure what that means).
http://cosmos.astro.uiuc.edu/cosmohome/workunit.php?wuid=87385
and
http://cosmos.astro.uiuc.edu/cosmohome/workunit.php?wuid=86180

Something is very wrong with the server (validator?) - it just seems to keep sending the same WUs over and over (even though they succeed). Eventually the 'maximum number of success results' figure (10) is reached and the whole lot gets 0 credit.


I've got a number or WUs heading that way... unless a match for my cpu should happen to come along which might give a matching result. I've got 3 or 4 results that have received 0.00 credit after 10 successful results yielded no match.
ID: 373 · Report as offensive     Reply Quote
Profile sysfried

Send message
Joined: 24 Jun 07
Posts: 114
Credit: 5,296,905
RAC: 0
Message 379 - Posted: 25 Jun 2007, 8:41:54 UTC - in response to Message 373.  



I've got a number or WUs heading that way... unless a match for my cpu should happen to come along which might give a matching result. I've got 3 or 4 results that have received 0.00 credit after 10 successful results yielded no match.


I'm now seeing WUs that are sent out 12 times without recieving results....
I've put my hosts on 'no new work' until this is fixed.

Regards,

Sysfried
ID: 379 · Report as offensive     Reply Quote
Profile Jord
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 15 Jun 07
Posts: 345
Credit: 50,500
RAC: 0
Message 384 - Posted: 25 Jun 2007, 11:22:56 UTC
Last modified: 25 Jun 2007, 11:23:09 UTC

I got 4 in a row which are deemed invalid for me. As to why? No clue.

WUID83403
WUID83300
WUID83250
WUID82986

All I can guess is that I claimed too much. Perhaps that's because I run with CPU throttle set to 80%. Then again, the rest of my results did get credit and ran at the same throttle.
ID: 384 · Report as offensive     Reply Quote
1 · 2 · 3 · 4 · Next

Forums : Technical Support : Invalid results