rpi_logo
started some new searches and updated the error limits
started some new searches and updated the error limits
log in

Advanced search

Message boards : News : started some new searches and updated the error limits

Author Message
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 56012 - Posted: 30 Oct 2012, 3:19:32 UTC

So while we're trying to figure out the problem with some of these workunits causing NAN/infinity fitnesses (and thus causing the boinc client to mark them as an error), I've started some new ones -- which should mean there are less errors floating around.

I've also updated out workunit templates so that after 1 error it should just trash a workunit (I hope). This means the erroneous workunits won't keep getting sent out to users further complicating the problem.

It's not the best solution, but hopefully it will improve things while we figure out what's wrong with the code/input files that's causing the issue for these searches.

--Travis
____________

GaryG
Avatar
Send message
Joined: 29 Aug 12
Posts: 31
Credit: 40,781,945
RAC: 0

Message 56013 - Posted: 30 Oct 2012, 3:20:48 UTC

Thanks for the update, much appreciated.

Link
Avatar
Send message
Joined: 19 Jul 10
Posts: 328
Credit: 16,291,849
RAC: 612

Message 56021 - Posted: 30 Oct 2012, 21:24:30 UTC - in response to Message 56012.

Looking good, only one "error while computing" today, my RAC is already again over 20,000 :).

That were the good news. As to the not-so-good part:

I've also updated out workunit templates so that after 1 error it should just trash a workunit (I hope). This means the erroneous workunits won't keep getting sent out to users further complicating the problem.

You might want to increase that to 2 (so actually 3): see this WU. I'm not sure how the whole thing is working, on this one new results (_3 & _4) were created after reaching the "too many errors" state, but I thought you might want to have a look at this one.


My system: Win7 x64, ATI HD3850 (not OpenCL capable, Catalyst 11.9) using the 64-bit CAL app v0.82, running 2 WUs at once.
____________
.

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 56024 - Posted: 31 Oct 2012, 2:42:12 UTC - in response to Message 56021.
Last modified: 31 Oct 2012, 2:42:24 UTC

Looking good, only one "error while computing" today, my RAC is already again over 20,000 :).

That were the good news. As to the not-so-good part:

I've also updated out workunit templates so that after 1 error it should just trash a workunit (I hope). This means the erroneous workunits won't keep getting sent out to users further complicating the problem.

You might want to increase that to 2 (so actually 3): see this WU. I'm not sure how the whole thing is working, on this one new results (_3 & _4) were created after reaching the "too many errors" state, but I thought you might want to have a look at this one.


My system: Win7 x64, ATI HD3850 (not OpenCL capable, Catalyst 11.9) using the 64-bit CAL app v0.82, running 2 WUs at once.



I don't think that's too much of an issue so long as those wingman who completed it actually got credit... and it looks like they did as far as I can tell.
____________

Link
Avatar
Send message
Joined: 19 Jul 10
Posts: 328
Credit: 16,291,849
RAC: 612

Message 56030 - Posted: 31 Oct 2012, 17:17:40 UTC - in response to Message 56024.

I don't think that's too much of an issue so long as those wingman who completed it actually got credit... and it looks like they did as far as I can tell.

Yeah, that's probably because my result was marked as invalid after it was compared with the other three results.

But see my post in Matthew's thread.
____________
.


Post to thread

Message boards : News : started some new searches and updated the error limits


Main page · Your account · Message boards


Copyright © 2018 AstroInformatics Group