Questions and Answers :
Wish list :
WU allocation - like to have preference to high RAC computers???
Message board moderation
Author | Message |
---|---|
Team Jolt Cola Send message Joined: 21 Aug 99 Posts: 31 Credit: 2,015,228 RAC: 0 |
Would it be possible (it may already be done) to make sure that when a WU is allocated to its first 3 users, that at least 3 of the HOSTS (or maybe users) are ones with high RAC - that way when a WU is given to a less dedicated cruncher at least 2 of them stand a chance of getting the result back to you in 2-3 days. Also if the "return by" date were shorter for those with less RAC then the higher producers would not be left with hundreds or thousands of WU's for which we have to wait several weeks before someone who only switches their pc on for 2-3 hours/day to return it before we get credit. Ideally, HIGH RAC HOSTS/USERS would only be allocated WU's which have 2 other HIGH RAC HOSTS/USERS - and LOW RAC users get put with other LOW RAC users. Then the high output people get RAC quickly and are not penalised with the rest. This would be fairly simple to implement I would have thought - you just put WU's into 2 (maybe more, but 2 is simple) groups (with relevant amounts of WU's of course) and allocate based on some notional RAC - say 500-100 or so. I ask this, as our RAC (and that of all the top producers) is dropping quickly - and suspect that pending credit is piling up (but we are unable to look) waiting for one slacker to do a WU. Or is it that the validator is having a bad time. Surely if there were 2 pools, one would have a very short time on the validator list - being the high RAC people - and the other would be all the rest who are clogging up the servers with hundreds of thousands of pending results. M. |
Keck_Komputers Send message Joined: 4 Jul 99 Posts: 1575 Credit: 4,152,111 RAC: 1 |
Under normal conditions a pool of pending credit gets built up so that some is being granted at about the same rate that new pending work is added to the pool. (Currently the backlog is because there was an error and alot of old work has to be revalidated.) I expect more delays are caused by workunit errors and long queues than by slower computers. So something like this is not needed for the reasons listed. I would like to see more classes of workunits though. I would like to see some with longer deadlines so that slower computers can be used and still make the deadline. John Keck BOINCing since 2002/12/08 |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.