Yep that would fix it, say total credit must be greater than the average.
Not a problem for any genuine computer.
The software probably comes from BOINIC so Eintein team can't fix it.
The software probably comes from BOINIC so Eintein team can't fix it.
Well, not exactly true... first, Bruce Allen of Einstein _is_ one of the "BOINC" developers, so he can make changes to the BOINC code. Obviously all the developers communicate, and changes aren't made by just anyone with a whim, but on something like this, I can't see anybody objecting.
Second, if it's something "project specific", the change could be made "locally". The downside to that is that the change must then be made _again_, any time an update was applied - but this is a fairly common thing, and can be automated to a degree. Look at the Einstein forum structure, for example; it's _much_ better than the "standard", with the non-functional "Q&A" section.
Plus, you can look on the developer's mailing list and see what's going on in the background at times. The archives are public, anyone can go read the month's traffic. AND, there's another thread here on this topic, in the help area, where Bruce has said what he's doing as he's done it. So I can tell you he discussed the pros/cons of running the cleanup script with others who are already doing it, then looked at the code himself to be sure, then ran it, and has set it up to run daily. Saw the same bogus computer at the top, and is currently verifying that no other project objects to him putting in a 'minimum' of 300 credits to qualify to be on the list in _any_ position. (A constant is much easier than comparing total to RAC, etc.)
The software probably comes from BOINIC so Eintein team can't fix it.
Well, not exactly true... first, Bruce Allen of Einstein _is_ one of the "BOINC" developers, so he can make changes to the BOINC code. Obviously all the developers communicate, and changes aren't made by just anyone with a whim, but on something like this, I can't see anybody objecting.
Second, if it's something "project specific", the change could be made "locally". The downside to that is that the change must then be made _again_, any time an update was applied - but this is a fairly common thing, and can be automated to a degree. Look at the Einstein forum structure, for example; it's _much_ better than the "standard", with the non-functional "Q&A" section.
Plus, you can look on the developer's mailing list and see what's going on in the background at times. The archives are public, anyone can go read the month's traffic. AND, there's another thread here on this topic, in the help area, where Bruce has said what he's doing as he's done it. So I can tell you he discussed the pros/cons of running the cleanup script with others who are already doing it, then looked at the code himself to be sure, then ran it, and has set it up to run daily. Saw the same bogus computer at the top, and is currently verifying that no other project objects to him putting in a 'minimum' of 300 credits to qualify to be on the list in _any_ position. (A constant is much easier than comparing total to RAC, etc.)
Bill, thanks very much for the 100% correct summary! [One missing step that entailed some extra delay was that I ALSO tested the update_stats script on the hidden Einstein@Home test project.]
Alan, what Bill has said is completely right. Generally speaking, when a problem appears, I would much rather fix it in the BOINC infrastructure rather than just on Einstein@Home. This is partly for altruistic reasons and partly for selfish ones. On the altruistic side, I would like to help other projects too. On the selfish side, it's a hassle to keep merging E@H specific changes into the baseline BOINC project code.
In any case, I am now running the update_stats program every 24 hours to decay the RAC values, and have also restricted the list of top hosts to those with at least 300 credits. The latter change has made in the BOINC base code, so should eventually appear in all BOINC projects.
All I meant that it was not entirely an Einstein issue but linked to BOINC so that a fix would involve both parties as you have shown.
And indeed some parties whilst of two were one.
And it came to pass that we don't have to look at obviously wrong statistics.
And Even though its decay rate was mathematically interesting.
All was good.
Thanks again.
Alan
PS
And thanks for the Altivec optimized clients too.
Sorry to be a pain but the top computer board is wrong again:-
498180 1 XJR-Maniac 20,773.04 3,658.61 GenuineIntel
x86 Family 6 Model 8 Stepping 10 1102MHz Microsoft Windows 2000
Professional Edition, Service Pack 3, (05.00.2195.00)
They don't appear in the Top Participants list, perhaps you could use the same algorithm you use there?
Yep that would fix it, say
)
Yep that would fix it, say total credit must be greater than the average.
Not a problem for any genuine computer.
The software probably comes from BOINIC so Eintein team can't fix it.
Alan
RE: The software probably
)
Well, not exactly true... first, Bruce Allen of Einstein _is_ one of the "BOINC" developers, so he can make changes to the BOINC code. Obviously all the developers communicate, and changes aren't made by just anyone with a whim, but on something like this, I can't see anybody objecting.
Second, if it's something "project specific", the change could be made "locally". The downside to that is that the change must then be made _again_, any time an update was applied - but this is a fairly common thing, and can be automated to a degree. Look at the Einstein forum structure, for example; it's _much_ better than the "standard", with the non-functional "Q&A" section.
Plus, you can look on the developer's mailing list and see what's going on in the background at times. The archives are public, anyone can go read the month's traffic. AND, there's another thread here on this topic, in the help area, where Bruce has said what he's doing as he's done it. So I can tell you he discussed the pros/cons of running the cleanup script with others who are already doing it, then looked at the code himself to be sure, then ran it, and has set it up to run daily. Saw the same bogus computer at the top, and is currently verifying that no other project objects to him putting in a 'minimum' of 300 credits to qualify to be on the list in _any_ position. (A constant is much easier than comparing total to RAC, etc.)
Thankyou very much for fixing
)
Thankyou very much for fixing it.
Apologies for my ignorance of BOINC and its develpers, and my spelling (BOINIC).
Alan
RE: RE: The software
)
Bill, thanks very much for the 100% correct summary! [One missing step that entailed some extra delay was that I ALSO tested the update_stats script on the hidden Einstein@Home test project.]
Alan, what Bill has said is completely right. Generally speaking, when a problem appears, I would much rather fix it in the BOINC infrastructure rather than just on Einstein@Home. This is partly for altruistic reasons and partly for selfish ones. On the altruistic side, I would like to help other projects too. On the selfish side, it's a hassle to keep merging E@H specific changes into the baseline BOINC project code.
In any case, I am now running the update_stats program every 24 hours to decay the RAC values, and have also restricted the list of top hosts to those with at least 300 credits. The latter change has made in the BOINC base code, so should eventually appear in all BOINC projects.
Bruce
Director, Einstein@Home
All I meant that it was not
)
All I meant that it was not entirely an Einstein issue but linked to BOINC so that a fix would involve both parties as you have shown.
And indeed some parties whilst of two were one.
And it came to pass that we don't have to look at obviously wrong statistics.
And Even though its decay rate was mathematically interesting.
All was good.
Thanks again.
Alan
PS
And thanks for the Altivec optimized clients too.
RE: And it came to pass
)
Absolutely. I am grateful for having this pointed out. I simply hadn't realized that it wasn't right until you and others spoke up!
Director, Einstein@Home
Sorry to be a pain but the
)
Sorry to be a pain but the top computer board is wrong again:-
498180 1 XJR-Maniac 20,773.04 3,658.61 GenuineIntel
x86 Family 6 Model 8 Stepping 10 1102MHz Microsoft Windows 2000
Professional Edition, Service Pack 3, (05.00.2195.00)
They don't appear in the Top Participants list, perhaps you could use the same algorithm you use there?
Alan