21)
Message boards :
Current tests :
New crediting system
(Message 2119)
Posted 16 Aug 2006 by dcdc Post: I think as long as overall it will average out to within a reasonable margin after running a few WUs then that's good enough for anyone. Do we need a reasonably large scale test of this or can the testing be done on Rosetta concurrent with the current system? Cherry-picking will not be possible because you will only know how many models you completed _after_ you crunched your WU. Cherry-picking will be possible unfortunatley, but it would take a bit of work to set up and I'm not going to say how ;) It will also be identifiable! |
22)
Message boards :
Current tests :
New crediting system
(Message 2106)
Posted 16 Aug 2006 by dcdc Post: As I haven't processed that much over @ the Rosetta Project for awhile I can't wait for the Fire Storm to hit the Project when & if it's done, I don't think the Dev's know what their letting themselves get into if thats implemented ... ;) I think you're probably right. The most vocal group might be the winners on this, but as FC says, this might be the best result for the project if fewer leave. I'd like to see what the effect on the data would be though as I expect it'd bring the top teams a bit closer together which'd be great for competition! |
23)
Message boards :
Current tests :
New crediting system
(Message 2099)
Posted 16 Aug 2006 by dcdc Post: We have the opportunity to back date this to make the credit system consistent and fair throughout. If you've crunched more than me, you'll have more credit than me. I have no Ralph credits but we're discussing Rosetta. [edit]i think there's a misunderstanding here - we're (or at least I'm) not taking about back-dating Ralph credits - this is about Rosetta credits[/edit]
I know it's annoying to loose credits for work done, and I don't know about any of the other projects, but I don't know of any lost any data on Rosetta. Also, I know it didn't happen from the start, but you should have been credited for any WU errors for the last four months or so. If you're being given credit for the work done I wouldn't consider that 'losing' credits though. You'd get the right credits for the amount of work completed. |
24)
Message boards :
Current tests :
New crediting system
(Message 2091)
Posted 16 Aug 2006 by dcdc Post: First off, I think it needs to be accepted that some people will stop crunching Rosetta because of this credit change, no matter how it's done. However, it is also the case that many people aren't currently crunching because of the credit system, and many aren't pushing what they can because of the lack of fair competition. Of course the drop-off will be quick and the gains will be gradual. However, if the situation isn't rectified then it will always be a contentious issue and I believe the project will suffer in the long run, so I'm sure it's the right thing to do for the project. I agree entirely with Ethan, and my position on the whole 'optimised' boinc manager situation is that it isn't in breach of the rules, but I consider it poor sportsmanship. Taking it to its logical extreme, if I add a magic Cell board to my computer that boosts my computer's SSE performance by 50x, Rosetta won't benefit in the slightest, but I'll get 50x more credits if using an optimised client. The credits aren't alligned to the work done. I'd like that to be an irellevant point on this topic, but I'm afraid its going to be the central theme. I do, however, appreciate what Zombie67 has said about people joining with one set of crediting rules, and then these being changed retro-actively. However, I can't sway myself from thinking that we have the opportunity to make the credit system fairly accurate based entirely on work done. I'm fairly sure everyone agrees credit should be based on work done. We have the opportunity to back date this to make the credit system consistent and fair throughout. If you've crunched more than me, you'll have more credit than me. If not, you won't. We know that's not the case at the moment, and I don't see how anyone can logically be against that. The right thing to do for the project is to keep as much crunching power as possible long term. Just a quick question, we are not going to have this split system (work credit and credit) at Rosetta? I agree - I don't think the stats sites would appreciate running two credit systems and it would just add to confusion. |
25)
Message boards :
Current tests :
New crediting system
(Message 2049)
Posted 15 Aug 2006 by dcdc Post: dcdc: I think the new credit system will be accepted much more willingly and quickly if the credits are backdated. If they could be calculated back to the start then that'd be best but if they can only go back to Feb then fair enough. If it's a lot of work to calculate then I'm sure we could take some of the burden from you on that front if you're willing to send one of us the data. Danny |
26)
Message boards :
Current tests :
New crediting system
(Message 2044)
Posted 15 Aug 2006 by dcdc Post: Hi everyone - I've just registered for Ralph, although I've not really had any problems with Rosetta in the past so i've never connected any computers up to it. I could drop a 12hr/day P3 1GHz on it in a few weeks if that's any use? I read the entire thread last night(!) and I think the new credit system is definitely along the right lines. I'm not sure about using Ralph to calculate a WU mean score though if the Ralph app differs from the Rosetta one (doesn't it produce debug info etc?). What about when you release new versions onto Ralph that take more or less time than the current Rosetta release? It might be wise to hand pick some known dependable machines for this task, whether on Ralph or Rosetta, and use those to calibrate the credits per decoy, rather than using the whole of Ralph as a test bed. They would just have to have a consistent hardware config (and preferably on 24/7 for a fast turn-around time), and you'd need to monitor/be informed if this changed. Would that take a lot of work? I guess you'd need a table of 'golden machines' that would be sent the jobs first, then the credit per decoy could be calculated before general release. Of course the 'golden' machines would be given whatever credit they request, so they'd have to be from trusted sources, but there are plenty of those available. I don't mind submitting a machine that will have a constant hardware config - i'm going to be putting a backup server in the loft at some point so it'd be fine to run that.
I've requested this over on the Rosetta forums as I'd like to have a look at this too (although this is probably a much better place to ask!). It seems to me it would probably be a fairly simple task to backdate the new credit model fairly accurately for all previous jobs, using computers which are known to have the same hardware/software config from the start, then calculating a credits/hour figure for each computer and then using the number of decoys/time taken to calculate the number of credits due for each type of WU decoy. I know there is a risk that some people will use this as fuel for more flaming, but I think we'll be able to get a good idea of whether the credits can be backdated reasonably accurately from this. Could a table showing the WU name, number of decoys produced, time taken and the benchmark score be released? I know it's gonna be a big file! Danny |
©2024 University of Washington
http://www.bakerlab.org