Daily quata of 8 results
2 May 2005 0:19:35 UTC
Topic 189094
(moderation:
I have a 3 day cache. One of my crunchers is down to 24 hours of work. When I connect, I get the brush from the server. "not sending work, daily quota of 8 results, yadda yadda". What's the Official explanation of this? Does my account have a limit on contributions to the project? My congratulations to all Einstein Crunchers. Pay your utility bills and feel warm.
Daily quata of 8 results
)
> I have a 3 day cache. One of my crunchers is down to 24 hours of work. When
> I connect, I get the brush from the server. "not sending work, daily quota of
> 8 results, yadda yadda". What's the Official explanation of this? Does my
> account have a limit on contributions to the project? My congratulations to
> all Einstein Crunchers. Pay your utility bills and feel warm.
>
Every user on every project has a daily quota. If you have WUs that are late, or have errored out, the quota goes down by one for each late or errored WU. The quota goes up by one for each successful on time result. The min quota is one, and the max is determined by the project. I know that CPDN is 4 / day. I do not remember what the max for any other project is.
BOINC WIKI
> Every user on every project
)
> Every user on every project has a daily quota. If you have WUs that are late,
> or have errored out, the quota goes down by one for each late or errored WU.
> The quota goes up by one for each successful on time result.
Quota DOUBLES for each successful result. See this entry on FAQ page.
Bruce
Director, Einstein@Home
I now have 4 computers with
)
I now have 4 computers with the quota of 8 results bug. These machines have been running since I signed on. I am not satisfied with the previous
"explanation". If Einstein is culling rigs that are outside the accuracy requirements of the project, I need to have a few more details. I have one rig that is only mildly overclocked running Boinc 4.25 that just went "quota 8 results" today. As we speak, I have 1.5 work units left on this machine. This rig has never posted any computation errors. I have a 4 day cache requirement, and I get diddly from the scheduler. 8 results means I do 4 WU per day and upload every other day. Something smells here. I am going to do a fresh install of XP and Boinc 4.36 on on of these rigs and see what kind of reception this gets from THE SCHEDULER. My guess is UWM needs to do some serious tweaking on the scheduler, else I need to fix my preferences. Advise please.
Just installed Boinc 4.37 on
)
Just installed Boinc 4.37 on one of my "Einstein quota 8's", LHC failed, Einstein said "Duhhhh". Went to Boikly for the "other project", and gee whiz, crunching as we speak. Something is not cool. Fix needed, I think. 8 results is dirt why am I limited by bad explanations?
I'm just curious as to why
)
I'm just curious as to why you're refering to this as a 'bug'? Everyone's maximum quota is 8 work units per day, per CPU, so you are getting the maximum allowable number of work units. I don't think you've been singled out in any way.
Digger: When I look at my
)
Digger: When I look at my preferences, in each of my computer profiles, this data element is not present for each computer. Some have it, some don't. 8 WU per day does not explain why I get stiffed for sending in 7/8 per two days per box. I'm sure this is not an issue for a pentium 233, but an A64 @2.5ghz? Excuse me! The data base is out of synch with the host file. I appreciate your response, but what is the value to the project to limit a box to 8 per day? Is the scheduler that lame?
Hiya tweakster, I can't
)
Hiya tweakster,
I can't answer your questions actually, I just wanted to put your mind at ease if you thought the system was picking on you specifically. We're all in the same boat so to speak. I didn't realize that the 8 per/day quota pertained to how many work units you could download in one day. I thought it meant how many per day you were allowed to cache. From what you are saying, the system is limiting you to a 2-day cache, when you would prefer to keep 3 days worth. I get it now. :)
At any rate, hope you find a resolution to your problem.
Dig
... From what you are saying,
)
...
From what you are saying, the system is limiting you to a 2-day cache
...
not quite.
Either you have one mean machine and are crunching more than 8 wu per cpu, in which case the cache will be less than one day and you will have some dead time each day unless you also run other projects,
Or (more likely), you crunch less than 8 wu per day, in which case the quota cannot affect the eventual size of your cache, it only affects how long you take to get to build up to a full cache. For example, crunch 5/day, and your cache grows at 3 wu/day.
Please also note that the max size of your cache is, by definition, bigger than the connection interval. The system tries to get the difference between your largest and smallest cache (high tide and low tide) to match the connection interval, but also gives you some more cache to act as a buffer. With most BOINC projects the cache is around 1.8x the interval, but with E@H most people find it to be around 2.5x to 3x. (This difference is connected with the fact that E@H underestimates the time a wu will take).
To get a cache of the size you want, start off with the connection interval around 1/4 of your preferred cache size. Wait till the cache stops growing, then increase the connection interval by 0.1 at a step till you get roughly what you want.
Notice too that the software tries to download and report wu in bunches: this is to reduce the number of database connections, which is one of the critical factors for performance. The larger your cache the more the software will try to take advantage of this - it's not a bug it is deliberate!
~~gravywavy
Tweakster, thank you for
)
Tweakster, thank you for running E@H. I'm sorry that you are having problems not getting enough work. Let me make some suggestions.
First off, if you have not done so, please read the FAQ where it describes how the maximum daily result quote is changed depending upon host behavior.
Second, could you please examine your 24 machines in the database? Unless you really own 24 machines, some of these could be merged. This would make it easier for me to study the behavior of them to understand what's going on.
From looking in the database, your machines take 6 cpu hours to do a workunit. So with the daily quota of 8 per day, if you have your machine running E@H 100% of the time, after about a week you can build up a cache of one week of workunits.
I see however that in the database many of the results assigned to your machines have 'timed out'. This may result in having the daily result quota lowered as described in the FAQ. So please try to understand why this is happening. When results time out this lowers the daily quota and might give rise to problems related to not getting enough work.
Cheers,
Bruce
Director, Einstein@Home
Bruce; Thanks for the reply.
)
Bruce; Thanks for the reply. In the last 2 months I have had to rebuild some of my machines due to motherboard failure or other component issues. That means any WU's associated with these machines would have gone delinquent. I have no ability to delete these machines, only merge them. This is a mine field I would rather not navigate. The vagaries of editing my computers is too daunting in its current configuration. I tried merging two once, and I think that's when the trouble started. Having said that, I have one machine that has in the last day gone "Quota 8" for no obvious reason. No configuration changes, no disabling overclock, in fact, it is my slowest box. It was a clean install of XP and SP2, Boinc 4.25, and running Norton SCS. It runs the Widows firewall along with Symatec and is behind a router and switch. I have had zero problems with this rig until "quota 8" out of the blue yesterday. I feel something is amiss with the way work unit requests are being allocated by the scheduler. I notice other crunchers are getting the dreaded "quota 8" denials also.
Regards; Tweakster