Advanced search

Message boards : Number crunching : Deadline extension

Author Message
Killer 69
Send message
Joined: 7 Jan 09
Posts: 3
Credit: 3,624,425
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwat
Message 28648 - Posted: 21 Feb 2013 | 13:03:51 UTC

Currently long runs have a pretty short deadline compared to the time, which it takes to process them causing many of them to fail to get processed in time due to computer not being on/computing all the time. Any possibility to extend that deadline, say to 10 days or so? Pretty frustrating to notice that computer has spent numerous hours crunching for nothing.

Profile microchip
Avatar
Send message
Joined: 4 Sep 11
Posts: 110
Credit: 326,102,587
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 28649 - Posted: 21 Feb 2013 | 13:34:51 UTC

This has been discussed before and I don't think they'll extend it.
____________

Team Belgium

werdwerdus
Send message
Joined: 15 Apr 10
Posts: 123
Credit: 1,004,473,861
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 28661 - Posted: 21 Feb 2013 | 17:48:18 UTC - in response to Message 28648.

Turn down your minimum and maximum additional work buffer to something like 0.05 or 0.1 days. That will keep BOINC from downloading a new work unit until the current one is almost finished.

They will not extend the deadlines, since each work unit, when completed, forms the beginning of the next work units (in a nutshell; it is a bit more complicated :P). They must come in soon after they are sent out so the queue can stay full and not waiting on work units that have been hoarded (either on purpose or by accident).
____________
XtremeSystems.org - #1 Team in GPUGrid

Dagorath
Send message
Joined: 16 Mar 11
Posts: 509
Credit: 179,005,236
RAC: 0
Level
Ile
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 28667 - Posted: 22 Feb 2013 | 6:03:42 UTC - in response to Message 28648.

You have 2 choices:

Subscribe to the short queue or setup a script that will wait until the GPUgrid task completes, uploads and reports before shutting down your PC.

Also, if other projects are pre-empting your GPUgrid task then you should set aside 1 core. That way the GPUgrid task will never get pre-empted and it will finish sooner.

____________
BOINC <<--- credit whores, pedants, alien hunters

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2705
Credit: 1,311,122,549
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 28707 - Posted: 23 Feb 2013 | 17:04:31 UTC

In addition to what the others said: I assume you're concerned about your machine with a Quadro K1000M. This GPU is based on GK107, the same chip as in GT640 and GTX650. this one is rather weak, but is even cut in half for that mobile Quadro (only 1 of 2 SMX blocks active). You shouldn't run any "Long runs" with it.. deactivate these in your profile and deactivate "if no work is available, accept work from other queues". This will mean that sometimes no "Short runs" tasks will be available, so you might want to set up a backup project with short WUs.

MrS
____________
Scanning for our furry friends since Jan 2002

Post to thread

Message boards : Number crunching : Deadline extension

//