Welcome to MilkyWay@home

might have found the error

Message boards : News : might have found the error
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 55768 - Posted: 13 Oct 2012, 23:40:07 UTC

It looks like the star file was misspecified for the *22_3s_free_1/2 and *22_3s_edge_1/2 searches. I've made a new fixed star file for *22_3s_free_3 and *22_3s_edge_3 searches, and shut down the old ones. Let me know how these are working.
ID: 55768 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile ritterm
Avatar

Send message
Joined: 16 Jun 08
Posts: 93
Credit: 366,882,323
RAC: 0
Message 55772 - Posted: 14 Oct 2012, 2:37:00 UTC - in response to Message 55768.  

...I've made a new fixed star file for *22_3s_free_3 and *22_3s_edge_3 searches...Let me know how these are working.

I've run through only about 30 of these these, but, so far, so good. Previously, my error rate had been about 10%.

ID: 55772 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
TLSI2000

Send message
Joined: 15 Mar 10
Posts: 17
Credit: 1,221,936,867
RAC: 0
Message 55773 - Posted: 14 Oct 2012, 2:58:27 UTC

A Follow-up...

After two hours of the version '3' , I have seen *zero* errors on them.

Still having a few errors on the version '1' and '2' WUs as those batches run their course.

It looks like the problem is solved.

Thx.
ID: 55773 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile ritterm
Avatar

Send message
Joined: 16 Jun 08
Posts: 93
Credit: 366,882,323
RAC: 0
Message 55774 - Posted: 14 Oct 2012, 3:20:10 UTC - in response to Message 55772.  
Last modified: 14 Oct 2012, 3:23:37 UTC

...so far, so good.

Or, maybe not so much... This result, 319700259, shows a failure similar to the earlier runs. :-(
ID: 55774 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 55775 - Posted: 14 Oct 2012, 5:40:20 UTC - in response to Message 55774.  

...so far, so good.

Or, maybe not so much... This result, 319700259, shows a failure similar to the earlier runs. :-(


Looks like the job completed successfully but for some reason the client marked it as a failure (it does have the output for the likelihood at the end). I wonder if this is due to the fact that one of the stream only likelihoods ended up being NAN.

I hope the errors like this are at least less frequent...
ID: 55775 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile dskagcommunity
Avatar

Send message
Joined: 26 Feb 11
Posts: 170
Credit: 205,557,553
RAC: 0
Message 55776 - Posted: 14 Oct 2012, 7:43:20 UTC

Had some errors overnight like this :

http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=319788399
DSKAG Austria Research Team: http://www.research.dskag.at



ID: 55776 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
TJ

Send message
Joined: 12 Aug 09
Posts: 262
Credit: 92,631,041
RAC: 0
Message 55779 - Posted: 14 Oct 2012, 10:57:13 UTC

For edge_1 and edge_3 and free_1 and free_3 still many errors (while computing) and even more validate error for same, however wingman seems to run them fine and get validated.
I have the newest driver 1.4.1741 with HD5870 and BOINC 7.0.28 on Win7 x64.
Greetings from,
TJ
ID: 55779 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
TJ

Send message
Joined: 12 Aug 09
Posts: 262
Credit: 92,631,041
RAC: 0
Message 55780 - Posted: 14 Oct 2012, 11:08:51 UTC
Last modified: 14 Oct 2012, 11:09:24 UTC

Could it be a graphics driver problem?
I run Albert@home once a while as well and they did okay since yesterday as there are many validate errors too with ati as I saw with wingman. Nvidia seems to be okay.
Greetings from,
TJ
ID: 55780 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
TLSI2000

Send message
Joined: 15 Mar 10
Posts: 17
Credit: 1,221,936,867
RAC: 0
Message 55782 - Posted: 14 Oct 2012, 11:23:25 UTC

And I have seen a number of errors on the version '3' WUs overnight as well.

These seem to have an #IND in the result as:

<stream_only_likelihood> -3.638176510600306 -10.877692835483177 -1.#IND00000000000 </stream_only_likelihood>
ID: 55782 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile ritterm
Avatar

Send message
Joined: 16 Jun 08
Posts: 93
Credit: 366,882,323
RAC: 0
Message 55783 - Posted: 14 Oct 2012, 11:35:25 UTC - in response to Message 55775.  
Last modified: 14 Oct 2012, 11:37:03 UTC

Travis said:
I hope the errors like this are at least less frequent...

A rough calculation with my host's results shows an error rate of about 2%-3% over the last 4 hours, or so. So, they do seem to be less frequent (for me).
ID: 55783 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Link
Avatar

Send message
Joined: 19 Jul 10
Posts: 621
Credit: 19,254,980
RAC: 1
Message 55796 - Posted: 14 Oct 2012, 21:32:51 UTC - in response to Message 55775.  

I hope the errors like this are at least less frequent...

Yes, errors are less frequent compared 22_3s_free_1 and 22_3s_edge_1 searches, today I had just 3 bad *free_3* WUs. So that's way better than those 11 edge/free_1 from yesterday.

What I have noticed is we have a lot of WUs which need 3 results before they be validated. Here some examples: 253847186, 253944881, 254009156, 253771061, 253995223, 253888155. That's just from 1 page of my results, which still contains many "completed, validation inconclusive", some of which will very probably join this club.
ID: 55796 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 55798 - Posted: 14 Oct 2012, 22:07:31 UTC - in response to Message 55796.  

I hope the errors like this are at least less frequent...

Yes, errors are less frequent compared 22_3s_free_1 and 22_3s_edge_1 searches, today I had just 3 bad *free_3* WUs. So that's way better than those 11 edge/free_1 from yesterday.

What I have noticed is we have a lot of WUs which need 3 results before they be validated. Here some examples: 253847186, 253944881, 254009156, 253771061, 253995223, 253888155. That's just from 1 page of my results, which still contains many "completed, validation inconclusive", some of which will very probably join this club.


I've upped the quorums for some workunits a little bit because I was noticing we were getting some weird results validated. So there's no problem there, just a little bit extra validation on our end.
ID: 55798 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : News : might have found the error

©2024 Astroinformatics Group