Join PrimeGrid
Returning Participants
Community
Leader Boards
Results
Other
drummers-lowrise
|
Message boards :
Sophie Germain Prime Search :
Minimum quorum
Author |
Message |
|
I have three hosts crunching SGS LLR. In one of them (intel i5 2500 running windows 7), the minimum quorum reported to all the tasks it gets is 1. The other two (an intel i3 350 M and a t7400) get SGS tasks with a minimum quorum of 2, which means I only get credit after a wingman finishes the same task.
Is this just a coincidence (if so it is a very weird one, considering the number of tasks crunched) or the minimum quorum is in any way CPU defined?
____________
676754^262144+1 is prime | |
|
|
Some of the LLR tasks have adaptive validation meaning if a host is clean enough it can be given tasks without a wingman.
Now on the other ones are you getting the work first or second? If you actually get the work second you could still be under the adaptive validation and being set as a wingman against others who have not proven themselves.
| |
|
|
On the others, I've got tasks as first and as second as well. All of them, as far as I can see, have two valid results, so I guess those two hosts are not considered "clean enough".
Thanks for the quick reply.
____________
676754^262144+1 is prime | |
|
mfbabb2 Volunteer tester
 Send message
Joined: 10 Oct 08 Posts: 510 ID: 30360 Credit: 14,746,967 RAC: 38,600
                     
|
Replicates are probably randomly sent out to "good" hosts as a quality assurance verification.
____________
Murphy (AtP)
| |
|
|
Well, I can see two WU's with a wingman in my "clean host" now.
I was the first to report work, but the second to get it. So, I believe Pooh Bear's explanation seems to fit better.
I'm only crunching a couple subprojects, so I wonder if the adaptative validation only applies to SGS LLR or to other LLR subprojects too (I always get wingmaned work in PSP and 321).
Any idea on how "clean enough" and "not so clean" (or "not proven clean") hosts are identified?
____________
676754^262144+1 is prime | |
|
|
I have only seen the adaptive work on SGS and PPS so far.
They do not state what is considered clean. Plus they will every so often have yours double checked just to make sure it's still on par.
| |
|
|
Adaptive replication is detailed here, including an exact definition of what is considered a "clean host": http://boinc.berkeley.edu/trac/wiki/AdaptiveReplication
At primegrid, AR is used on 3 subprojects - PPS LLR, Sophie Germain LLR and the Riesel Problem LLR. All other LLR subprojects (321, etc) do not use this. i.e. these other workunits will require wingmen before credit is granted.
Regards,
Darryl
____________
| |
|
mfbabb2 Volunteer tester
 Send message
Joined: 10 Oct 08 Posts: 510 ID: 30360 Credit: 14,746,967 RAC: 38,600
                     
|
Of course, if a prime is found, it is then checked by multiple people/programs.
;-)
____________
Murphy (AtP)
| |
|
Message boards :
Sophie Germain Prime Search :
Minimum quorum |