Advanced search

Message boards : Wish list : sort of ECC support

Post to thread Subscribe Sort
Author Message
Profile X-Files 27
Avatar
Send message
Joined: 11 Oct 08
Posts: 95
Credit: 68,023,693
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15621 - Posted: 5 Mar 2010 | 16:11:38 UTC

Can ECC idea be incorporated in the app? Like for example the driver has crashed and has recovered, instead of throwing an error, pause for awhile then resume after sometime.
____________

ID: 15621 | Rating: 0 | rate: Rate + / Rate - Report as offensive    Reply Quote
Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15622 - Posted: 5 Mar 2010 | 17:38:59 UTC - in response to Message 15621.

We proposed that to BOINC at the start of GPUs, but finally they decided to keep the current policy.

gdf

ID: 15622 | Rating: 0 | rate: Rate + / Rate - Report as offensive    Reply Quote
Profile X-Files 27
Avatar
Send message
Joined: 11 Oct 08
Posts: 95
Credit: 68,023,693
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15626 - Posted: 6 Mar 2010 | 0:39:09 UTC - in response to Message 15622.

We proposed that to BOINC at the start of GPUs, but finally they decided to keep the current policy.

gdf

just noticed it today:
Fix - Recover from crashes in Nvidia and ATI GPU functions (Linux, Mac)

Is this true? how about windows, when will it be?
____________

ID: 15626 | Rating: 0 | rate: Rate + / Rate - Report as offensive    Reply Quote
Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15665 - Posted: 10 Mar 2010 | 12:15:50 UTC - in response to Message 15626.

For anyone experiencing GPUGrid task failures, now wold be a good time to update Boinc!

ID: 15665 | Rating: 0 | rate: Rate + / Rate - Report as offensive    Reply Quote
Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15674 - Posted: 10 Mar 2010 | 16:15:45 UTC - in response to Message 15626.

We proposed that to BOINC at the start of GPUs, but finally they decided to keep the current policy.

gdf

just noticed it today:
Fix - Recover from crashes in Nvidia and ATI GPU functions (Linux, Mac)

Is this true? how about windows, when will it be?



I don't thin that this is what you asked. But I am not sure.

gdf

ID: 15674 | Rating: 0 | rate: Rate + / Rate - Report as offensive    Reply Quote
Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15679 - Posted: 10 Mar 2010 | 19:51:45 UTC - in response to Message 15674.

Could applications be developed to incorporate a feature whereby if a failure occurs (which could be caused by just about anything) Boinc will first try to resumb from a time check point, and should that fail perhaps close Boinc and re-open Boinc, before erroring out? This might remove lots of random errors, and would bypass the need to rely on Boinc code to incorporate this or any other feature.

PS. I think it is important that GPUGrid takes the lead with such possible improvements, rather than relying on Boinc (which now has a gluttony of CPU projects to pander to). Your recent performance improvements came from your own endeavour. Keep it up.

ID: 15679 | Rating: 0 | rate: Rate + / Rate - Report as offensive    Reply Quote
Post to thread

Message boards : Wish list : sort of ECC support

//