Advanced search

Message boards : Number crunching : Working Practice - Corrupted Models

Post to thread Subscribe Sort
Author Message
Profile Zydor
Send message
Joined: 8 Feb 09
Posts: 252
Credit: 1,309,451
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 6715 - Posted: 17 Feb 2009 | 12:25:01 UTC

I had a hassle with one of my PC's over the last couple of days, that resulted in my aborting a model as there was no way I would meet the deadline, and two falling over with computation error. Got the beast back on track again now.

I am relatively new to GPUGRID, therefore greatful for advice. Its usual with most BOINC apps to give partial credits for time completed in cases where models fall over or end up with a corrupted result, does that happen here? Or is it a case of no result, no credits. There are merits in both approach's, am just curious to know what GPUGRID does in these scenarios.

ID: 6715 | Rating: 0 | rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 6723 - Posted: 17 Feb 2009 | 17:40:30 UTC - in response to Message 6715.

I had a hassle with one of my PC's over the last couple of days, that resulted in my aborting a model as there was no way I would meet the deadline, and two falling over with computation error. Got the beast back on track again now.

I am relatively new to GPUGRID, therefore greatful for advice. Its usual with most BOINC apps to give partial credits for time completed in cases where models fall over or end up with a corrupted result, does that happen here? Or is it a case of no result, no credits. There are merits in both approach's, am just curious to know what GPUGRID does in these scenarios.


In that the models take (in general) less than a day on even the slowest recommended cards the award is only for successful models.

On a project like CPDN where weeks are required the other award system makes more sense.

ID: 6723 | Rating: 0 | rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Zydor
Send message
Joined: 8 Feb 09
Posts: 252
Credit: 1,309,451
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 6735 - Posted: 17 Feb 2009 | 22:32:49 UTC - in response to Message 6723.

Fair enough - many thanks :)

ID: 6735 | Rating: 0 | rate: Rate + / Rate - Report as offensive    Reply Quote
Post to thread

Message boards : Number crunching : Working Practice - Corrupted Models

//