Simpler deadline

Message boards : News : Simpler deadline

Author	Message
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 Scientific publications	Message 21991 - Posted: 10 Sep 2011 \| 6:47:02 UTC
	Since a few days, the deadline for the results is fixed at 5 days.There is not anymore any special mechanism for replication of results as before. This change should help people with slow cards to complete workunits. Make sure you select the right application (short 3h or long 12h) for your card. gdf
	ID: 21991 \| Rating: 0 \| rate: / Reply Quote

Retvari Zoltan Send message Joined: 20 Jan 09 Posts: 2352 Credit: 16,297,596,199 RAC: 3,912,596 Level Scientific publications	Message 21992 - Posted: 10 Sep 2011 \| 9:09:52 UTC - in response to Message 21991. Last modified: 10 Sep 2011 \| 9:10:14 UTC
	I have mixed feelings about this. While it's a good idea to have only one deadline, maybe you'll have to increase the number of workunits in the queue to avoid outages due to longer return times. We'll see. A special mechanism would be needed to avoid sending long workunits to slow hosts.
	ID: 21992 \| Rating: 0 \| rate: / Reply Quote

skgiven Volunteer moderator Volunteer tester Send message Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level Scientific publications	Message 22080 - Posted: 12 Sep 2011 \| 22:43:21 UTC - in response to Message 21992.
	The deadline has been 5days for years. The difference is that GPUGrid is no longer automatically resending tasks after 3 days; which previously negated contributions from 3days+ crunchers, despite the fact that they previously did get credit. While I respect the facilitation of a greater number of cards and crunchers, I am concerned that this encourages slowness in returning tasks and the use of non-recommended cards that may very soon (4.1; CC1.3+ only) be unusable. My biggest concern however is the consequence of aborted or failed tasks (now that there is no automatic mechanism for reissuing such tasks). So crunchers, please stick to the recommendations (low cache, report tasks immediately), and please do not abort tasks, so long as they will finish within 5days. If you are using a low to mid-range card, or do not contribute frequently (say 12h+ per day) stick to the normal/short tasks.
	ID: 22080 \| Rating: 0 \| rate: / Reply Quote

Dagorath Send message Joined: 16 Mar 11 Posts: 509 Credit: 179,005,236 RAC: 0 Level Scientific publications	Message 22082 - Posted: 12 Sep 2011 \| 23:30:15 UTC - in response to Message 22080.
	My biggest concern however is the consequence of aborted or failed tasks (now that there is no automatic mechanism for reissuing such tasks). I had this task fail and it was resent so I think he meant there is no longer a special mechanism to resend after 3 days. The standard resend mechanism is still operational, it seems.
	ID: 22082 \| Rating: 0 \| rate: / Reply Quote

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 Scientific publications	Message 22085 - Posted: 13 Sep 2011 \| 6:57:15 UTC - in response to Message 22082.
	Yes, there is no longer the special mechanism to resend after 3 days, but the other standard mechanisms are still in place. gdf
	ID: 22085 \| Rating: 0 \| rate: / Reply Quote

microchip Send message Joined: 4 Sep 11 Posts: 110 Credit: 326,102,587 RAC: 0 Level Scientific publications	Message 22105 - Posted: 14 Sep 2011 \| 15:40:59 UTC - in response to Message 21991.
	Since a few days, the deadline for the results is fixed at 5 days.There is not anymore any special mechanism for replication of results as before. This change should help people with slow cards to complete workunits. Make sure you select the right application (short 3h or long 12h) for your card. gdf Hi, excuse this newbie the question, but I run a GT 440 here and my very first task from GPUGRID is reported by boinc as needing 27 hours to compute. You mention to select the right application but where can I do that? thanks :)
	ID: 22105 \| Rating: 0 \| rate: / Reply Quote

Retvari Zoltan Send message Joined: 20 Jan 09 Posts: 2352 Credit: 16,297,596,199 RAC: 3,912,596 Level Scientific publications	Message 22106 - Posted: 14 Sep 2011 \| 16:07:21 UTC - in response to Message 22105. Last modified: 14 Sep 2011 \| 16:07:54 UTC
	Hi, excuse this newbie the question, but I run a GT 440 here and my very first task from GPUGRID is reported by boinc as needing 27 hours to compute. You mention to select the right application but where can I do that? thanks :) Click on "Your Account" Look for the settings section, in the last line "Preferences for this project" click on "GPUGrid settings" Click on "GPUGrid settings" Uncheck the checkbox beside "ACEMD for long runs (8-12 hours on fastest GPU)" Click on save settings.
	ID: 22106 \| Rating: 0 \| rate: / Reply Quote

microchip Send message Joined: 4 Sep 11 Posts: 110 Credit: 326,102,587 RAC: 0 Level Scientific publications	Message 22107 - Posted: 14 Sep 2011 \| 17:34:03 UTC - in response to Message 22106.
	Hi, excuse this newbie the question, but I run a GT 440 here and my very first task from GPUGRID is reported by boinc as needing 27 hours to compute. You mention to select the right application but where can I do that? thanks :) Click on "Your Account" Look for the settings section, in the last line "Preferences for this project" click on "GPUGrid settings" Click on "GPUGrid settings" Uncheck the checkbox beside "ACEMD for long runs (8-12 hours on fastest GPU)" Click on save settings. Thanks :D
	ID: 22107 \| Rating: 0 \| rate: / Reply Quote

Post to thread

Message boards : News : Simpler deadline

	About	Science	Volunteers	Performance	Forum	Join us	Donate