| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Discovering Dengue Drugs - Together
Thread: Points attribution for crunching |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 31
|
|
| Author |
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
KerSamson,
----------------------------------------The 64 bit integer is the fast part of the XP64 - 64 Bit Client Benchmark (whetstone+dhrystone) / 480 = hourly claim. We do foremost 32bit floating point science. Thus if you see big differences, it is from these mostly "random victims" 64 meets 32 bit pairings. If 2 64 bit clients meet in quorum, you'll see/saw a sudden high claim. The averaging rules being introduced to eliminate these consistent underclaimers will smooth things further, mostly for the high claimers (and those always claiming too low are not interested anyhow, it seems) What's the OS on the Q6600... is that the XP64? My Q6600/Vista 32 bit gets nominally it's credit claims too.
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1684 Status: Offline Project Badges: 
|
Hello Sekerob,
----------------------------------------the Q6600 system is running Win XP SP2 32 bits with Boinc 32 bits. The variation between claimed and granted credits is very low and in the most cases the granted credit is a little bit higher than claimed. Cheers, |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Thanks, clear as a bell.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges: 
|
The validators are now using the new logic as described above. This logic applies only to 2 quorum workunits (so only FAAH, DDDT and HCC).
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello!
I thought I had understood the principles for points attribution, but now I found this for one of my machines: Project Name: Discovering Dengue Drugs - Together Created: 03/17/2008 13:32:37 Name: dddt0501c0152_100347 Minimum Quorum: 2 Initial Replication: 2 Result Name Status Sent Time Time Due / Return Time CPU Time (hours) Claimed/ Granted BOINC Credit dddt0501c0152_ 100347_ 0-- Valid 03/18/2008 17:18:17 03/19/2008 23:14:36 23.48 162.4 / 96.1 dddt0501c0152_ 100347_ 1-- Valid 03/18/2008 17:17:09 04/01/2008 08:15:18 21.12 105.0 / 96.1 <--mine Is this a result of the new logic? I thought even with the new logic the granted credits should be somewhere in the range between the lower and the higher claimed credit, and not outside of that range. Could someone explain? Best regards Thorsten |
|||
|
||||
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
Yes - this is due to the new logic. Since the claimed credits were so widely apart, it computed your credit by average historical credit per cpu time * cpu_time for the result for both results.
In this particular case, the historical average * cpu time was less then what both of you claimed so the credit awarded was lower then what either of you claimed. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
In this case the new logic is worse than the old.
At least I am not the one who was hit harder..... But I will keep an eye on my other results. |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Hi Thorsten,
----------------------------------------Maybe yes, maybe no, but at least the many who consistently heavily under-claim do not cause you or anyone be paired down big time as was happening. What the solace is, that if you'd been coupled to a 90ish claiming quorum partner, you'd not even known and shrugged it off as normal. So, yes keep watching and reporting if anything seriously deviating happens, for only then will we be able to inform WCG of flaws in the logic. cheers
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Apr 1, 2008 4:56:51 PM] |
||
|
|
zombie67 [MM]
Senior Cruncher USA Joined: May 26, 2006 Post Count: 228 Status: Offline Project Badges: 
|
As a result, we are going to change how the 2nd part of the process works. Instead of selecting the credit that is closest to its history, we will average the recent average history's for the two computers. We have been simulating the impact of this for the past couple of days and it turns out that in a strong majority of cases the result cpu time * host recent average credit per cpu second is actually quite consistent between different computers even if their claimed credit are further apart. This is what we had hoped to see and as a result we will start to use this policy in the near future. Has this change been put back into BOINC code yet? If not, can it please be? According to the description here: http://boinc.berkeley.edu/trac/wiki/ValidationSimple ...it does not look like it has. There is at least one other project also using "two_credit()", and they are running into the same problem you discovered. Thanks!  [Edit 1 times, last edit by zombie67 at Mar 14, 2009 6:06:43 PM] |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
How about that, your wish it WCG's command:
----------------------------------------David 16 Mar 2009 3047 - validator utilities: improvements to two_credit() 3048 (used for deciding how much credit to grant 3049 when there are 2 valid results). 3050 From Kevin Reed 3051 3052 sched/ 3053 validate_util.cpp
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
|