| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 11
|
|
| Author |
|
|
sk..
Master Cruncher http://s17.rimg.info/ccb5d62bd3e856cc0d1df9b0ee2f7f6a.gif Joined: Mar 22, 2007 Post Count: 2324 Status: Offline Project Badges:
|
Noticed my Boinc Combined Score fell by about 25 places in a day, so I looked into it and found that some participants in Aqua managed to get enormous credit. For example, 120M credits on a Celeron, in one day!
On the Aqua web site I found a message blaming a bug in the Boinc server software, "We are dealing with multiple, unrelated problems. The server config files have been fully restored. Then there was the bad XML format problem, caused by a bug in BOINC server code. The credit problem we see now first appeared about a month ago. This problem is not caused by TEST workunits, but with BOINC server software. There is a bug somewhere in the server code. We usually use the latest version of BOINC's server software, so we end up having to deal with such problems (that is not to say that we ourselves haven't caused problems in the past). All we can do is to hope that BOINC developers find the cause of the problem soon". http://aqua.dwavesys.com/forum_thread.php?id=835 Heads up team. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
As if WCG has a history of jumping in feet first, last looked still on version 6.01 as core. What they're on over there I don't know, but suspect there to be something in their designer water [if there is a pun, it was intended ;O]
--//-- |
||
|
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges:
|
And it looks like the problem is still not fixed....at least in the Combined Score.
-------------------------------------------------------------------------------- [Edit 1 times, last edit by BladeD at Jul 11, 2011 2:06:09 PM] |
||
|
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges:
|
Any idea when the stats will be fixed?
----------------------------------------From Aqua... I apprised them about this before the rollback but of course they didn't listen. Now the stats are stuck in the erroneous state. They need to do an export to get things back to normal. These stats used to be 10 postitions higher than the one in my sig.. ![]() |
||
|
|
Dataman
Ace Cruncher Joined: Nov 16, 2004 Post Count: 4865 Status: Offline Project Badges:
|
"2011-07-14: AQUA@home - User, Host, and Team credit were rolled back
----------------------------------------I just set the total_credit, expavg_credit, and expavg_time values for users, hosts, and teams to be the same as those of 19-June-2011. I'll be releasing a few hundred FP and TEST workunits to test the credit system. Please note that BOINC's CreditNew is at work now, and the amount of granted credit will not be the same as before." ![]() |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
This is the bit of code that we are currently working through our testing for updating the servers. It is a bit challenging to test unfortunately since our test computers tend to be relatively homogenous and we only have a few gpu capable machines. We are trying to see if there is a way that we can update the server code and utilize the existing credit schemes while allowing the new code to compute credit awards based on the new policies and allow to to take some deep looks at how it is behaving and what the impact will be from typical problems that appear from time to time (such as wrong fpops estimates, changing workunit sizes, etc).
We have also done some non-standard things with our credit granting schemes - in particular the 'per unit' credit that we award on projects such as hpf2 or rice or projects with caps on runtime such as cep2 and hcmd2. These don't fit well with these new policies. As a result, we are reviewing how we are doing things and how the code operates and figuring out the best way to move forward. Unfortunately this takes some time. I do like the underlying philosophy of the new credit code (i.e. using a rolling average over many workunits to compute the efficiency of a particular computer and each app version to derive adjustments to determine the credit award based on time spent running). However, we have to work through the details. |
||
|
|
Dataman
Ace Cruncher Joined: Nov 16, 2004 Post Count: 4865 Status: Offline Project Badges:
|
Thanks much for this information, Kevin.
----------------------------------------![]() ![]() |
||
|
|
sk..
Master Cruncher http://s17.rimg.info/ccb5d62bd3e856cc0d1df9b0ee2f7f6a.gif Joined: Mar 22, 2007 Post Count: 2324 Status: Offline Project Badges:
|
I strongly agree with the underlying philosophy and respect your efforts to move forward in the correct direction.
Between just a few crunchers here I know there are a variety of GPU's, spanning several generations of both main types. Should you want trial results just ask; I'm sure the community could quickly put together a good range of GPU's. , |
||
|
|
Dataman
Ace Cruncher Joined: Nov 16, 2004 Post Count: 4865 Status: Offline Project Badges:
|
I strongly agree with the underlying philosophy and respect your efforts to move forward in the correct direction. Between just a few crunchers here I know there are a variety of GPU's, spanning several generations of both main types. Should you want trial results just ask; I'm sure the community could quickly put together a good range of GPU's. ,I agree also. I have 12 Nvidia's at your disposal. ![]() ![]() |
||
|
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges:
|
Any idea when the stats will be fixed? From Aqua... I apprised them about this before the rollback but of course they didn't listen. Now the stats are stuck in the erroneous state. They need to do an export to get things back to normal. These stats used to be 10 postitions higher than the one in my sig.. ![]() Looks like it's fixed! ![]() |
||
|
|
|