Panic Mode On (23) Server problems

Message boards : Number crunching : Panic Mode On (23) Server problems
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 10 · 11 · 12 · 13

AuthorMessage
Bob Giel
Volunteer tester

Send message
Joined: 11 Jan 04
Posts: 76
Credit: 5,419,128
RAC: 0
United States
Message 930632 - Posted: 3 Sep 2009, 14:56:58 UTC - in response to Message 930625.  

Never delete your results because the uploads don't go!! Since Monday, I've had numberous work units that said "uploading". They went sometime early this morning. You have to trust the application. BOINC knows what's there and when the servers free up will take care of the uploads. There's also times when downloads freeze for a bit. I hope you don't delete those?
ID: 930632 · Report as offensive
Zebra3
Avatar

Send message
Joined: 22 Oct 01
Posts: 186
Credit: 13,658,148
RAC: 0
Canada
Message 930634 - Posted: 3 Sep 2009, 15:09:28 UTC - in response to Message 930632.  
Last modified: 3 Sep 2009, 15:10:31 UTC

Never delete your results because the uploads don't go!! Since Monday, I've had numberous work units that said "uploading". They went sometime early this morning. You have to trust the application. BOINC knows what's there and when the servers free up will take care of the uploads. There's also times when downloads freeze for a bit. I hope you don't delete those?



I believe that what Sutaru's is doing is reducing the pending uploads down below the CPUs x2 so that the schedular will give him a refill of his CUDA cache. My policy is if I have crunched the WU I will report it not abort it for any reason. Isn't that why other GPU projects are out there? I ran 5 long GPU WU's and several CPU WU's on my backup projects while I was waiting for Seti work. Some crunchers JUST want to do S@h work and that is fine with me but aborting finished WU's isn't right in my mind but that's just my opinion.

Cheers
http://www.novascotia.com
ID: 930634 · Report as offensive
Profile Dirk Sadowski
Volunteer tester

Send message
Joined: 6 Apr 07
Posts: 7105
Credit: 147,663,825
RAC: 5
Germany
Message 930636 - Posted: 3 Sep 2009, 15:48:56 UTC
Last modified: 3 Sep 2009, 15:55:14 UTC


Yes, sorry that I need to do it - but, why is BOINC unfair to GPU cruncher..?

I needed to delete the ULs down to 8, that BOINC can ask again for new work.
The GPU cruncher idled because of no work, since ~ 2 days or something..


'increase '> CPUs x2' in UL for work request' -> [http://setiathome.berkeley.edu/forum_thread.php?id=54711]

[ EDIT: Posted 18 Jul 2009 16:48:10 UTC]


ID: 930636 · Report as offensive
Iztok s52d (and friends)

Send message
Joined: 12 Jan 01
Posts: 136
Credit: 393,469,375
RAC: 116
Slovenia
Message 930645 - Posted: 3 Sep 2009, 16:52:10 UTC - in response to Message 930636.  


Yes, sorry that I need to do it - but, why is BOINC unfair to GPU cruncher..?

I needed to delete the ULs down to 8, that BOINC can ask again for new work.
The GPU cruncher idled because of no work, since ~ 2 days or something..



Once I was considering it, never done it.
But I am shure we can take WU from client_state.xml and save all relevant data,
when normal upload works, just add it to .xml file and upload.
Takes some time to study, but it shall work.
30 WUs is 30 WUs!

BR
Iztok


ID: 930645 · Report as offensive
1mp0£173
Volunteer tester

Send message
Joined: 3 Apr 99
Posts: 8423
Credit: 356,897
RAC: 0
United States
Message 930656 - Posted: 3 Sep 2009, 18:16:22 UTC - in response to Message 930617.  

All the work I got this morning came from a Sept and a Dec tape and they dont show at all on the splitter but I am sure someone will be able to inform me on the reason...my guess is an offline tape from storage?

Most likely reissues.
ID: 930656 · Report as offensive
1mp0£173
Volunteer tester

Send message
Joined: 3 Apr 99
Posts: 8423
Credit: 356,897
RAC: 0
United States
Message 930658 - Posted: 3 Sep 2009, 18:19:11 UTC - in response to Message 930636.  

Why is BOINC unfair to GPU cruncher..?

The word "unfair" seems to suggest that BOINC discriminates or that this is a design decision.

It's far more likely that this is an area where the BOINC work fetch algorithm doesn't work as well as it could.

I don't know if the BOINC project needs feedback on this, or if they're fully up to speed, but I seem to remember reading about changes that should make it work better.

Of course, if work is tight, it's possible that there just isn't enough.

ID: 930658 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 930659 - Posted: 3 Sep 2009, 18:26:42 UTC - in response to Message 930656.  

All the work I got this morning came from a Sept and a Dec tape and they dont show at all on the splitter but I am sure someone will be able to inform me on the reason...my guess is an offline tape from storage?

Most likely reissues.

No - split today:

14se08ae.459.387046.9.10.95
04dc08ag.1364.58447.9.10.252
ID: 930659 · Report as offensive
Ingleside
Volunteer developer

Send message
Joined: 4 Feb 03
Posts: 1546
Credit: 15,832,022
RAC: 13
Norway
Message 930684 - Posted: 3 Sep 2009, 21:08:10 UTC - in response to Message 930636.  

Yes, sorry that I need to do it - but, why is BOINC unfair to GPU cruncher..?

BOINC v6.10.x uses the max of ncpus, copros_cuda-count and coproc_ati-count, so it doesn't matter if it's gpu or cpu.

"I make so many mistakes. But then just think of all the mistakes I don't make, although I might."
ID: 930684 · Report as offensive
Previous · 1 . . . 10 · 11 · 12 · 13

Message boards : Number crunching : Panic Mode On (23) Server problems


 
©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.