Work Unit Time issues

Message boards : Graphics cards (GPUs) : Work Unit Time issues

Author	Message
skgiven Volunteer moderator Volunteer tester Send message Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level Scientific publications	Message 9148 - Posted: 30 Apr 2009 \| 19:16:42 UTC
	I am in the unfortunate position of having an 8600 GT Card. Typically the WU’s take about 60h to run on my system. The latest WU I started is due to be returned in 5 days! So, to complete it I will need to have my system on for more than 12h per day for the next 5 days. Would it not make more sense to have work units that can be completed in say, 15 days? The amount of successful tasks would surely rise significantly. I know some people have £1500 graphics cards, but I don’t, I don’t want one, and I am not going to get one. ____________ FAQ's HOW TO: - Opt out of Beta Tests - Ask for Help
	ID: 9148 \| Rating: 0 \| rate: / Reply Quote

ExtraTerrestrial Apes Volunteer moderator Volunteer tester Send message Joined: 17 Aug 08 Posts: 2705 Credit: 1,311,122,549 RAC: 0 Level Scientific publications	Message 9157 - Posted: 30 Apr 2009 \| 20:17:24 UTC - in response to Message 9148.
	The project needs results back quickly to base their future WUs on these, that's why the deadlines are rather short. The project knows that slower cards are struggling with this demand, that's why at least 50 shaders are recommended. Your card has only 32 of them, but is nevertheless still able to return WUs within the deadline.. if things don't go wrong. Take a look here, the recent monster WUs were not intended to be this big. MrS ____________ Scanning for our furry friends since Jan 2002
	ID: 9157 \| Rating: 0 \| rate: / Reply Quote

robertmiles Send message Joined: 16 Apr 09 Posts: 503 Credit: 757,780,562 RAC: 351,841 Level Scientific publications	Message 9221 - Posted: 2 May 2009 \| 20:53:22 UTC Last modified: 2 May 2009 \| 20:54:39 UTC
	Is the project able to consider the number of shaders available, and the recent average time the machine has been available, in deciding which machines to send those oversize workunits with short deadlines to? Does it offer a limit on the upload speed so that the internet link will be usable for other purposes during the upload?
	ID: 9221 \| Rating: 0 \| rate: / Reply Quote

Paul D. Buck Send message Joined: 9 Jun 08 Posts: 1050 Credit: 37,321,185 RAC: 0 Level Scientific publications	Message 9235 - Posted: 3 May 2009 \| 6:33:26 UTC
	All of those would be features of the BOINC Client and not at all under the perview of the project. Thought they could ask for them from the developers. We, the insane, have been asking that they consider making the resource scheduler smarter from the perspective of GPUs and I noted back when they were discussing features to be in 6.6.x that the assumption that the pool of GPUs would be all of the same class was unrealistic. Of course that line of questioning was ignored. That seems to be a historical constant, I keep asking questions that the developers do not want to talk about much less think about.
	ID: 9235 \| Rating: 0 \| rate: / Reply Quote

ExtraTerrestrial Apes Volunteer moderator Volunteer tester Send message Joined: 17 Aug 08 Posts: 2705 Credit: 1,311,122,549 RAC: 0 Level Scientific publications	Message 9250 - Posted: 3 May 2009 \| 12:05:59 UTC - in response to Message 9221.
	I assume the tactic is not to take "rush" action and let this mistake correct itself, i.e. let them run until completion and try not to generate such large ones again. MrS ____________ Scanning for our furry friends since Jan 2002
	ID: 9250 \| Rating: 0 \| rate: / Reply Quote

Post to thread

Message boards : Graphics cards (GPUs) : Work Unit Time issues

	About	Science	Volunteers	Performance	Forum	Join us	Donate