| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: The Clean Energy Project - Phase 2 Forum
Thread: how are calculated points ? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 12
|
|
| Author |
|
|
boulmontjj
Senior Cruncher France Joined: Nov 17, 2004 Post Count: 317 Status: Offline Project Badges: 
|
Sometime, something is strange with this research ; the points calculation.
----------------------------------------Let have a look to one WU i returned : E200372_ 634_ A.24.C19H14N2OSSi.78.2.set1d06_ 1-- 619 Valide 22/09/10 01:14:07 25/09/10 04:59:45 5,54 103,0 / 168,8 E200372_ 634_ A.24.C19H14N2OSSi.78.2.set1d06_ 0-- 619 Valide 22/09/10 00:49:13 24/09/10 21:21:10 11,16 117,3 / 84,4 The first line is a member that return the WU the 25th after 5,54 hours crunch and 103 points claimed. He obtained 168,8 points ! I returned the same WU the 24th (before him) after 11,16 hours crunch (more than him) and only claimed 117,3 points. And i obtained just 84,4 poor points !!!  What does that mean ? What is the rule to calculate points ? |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Look inside the 2 logs how many jobs of the 16 were really computed. For 99 out of 100 my client receives exact middle of quorum and logs are identical (my quad loves to do CEP2 under Linux). Very sometimes my client skips less jobs than the wingman and then some proration rule kicks in, similar to RICE and HCMD2, counting the seeds/positions.
----------------------------------------E200338_ 463_ A.23.C20H14OSi2.25.3.set1d06_ 0-- 1292373 Valid 12-9-10 21:21:49 19-9-10 16:14:55 6.10 109.9 / 117.3 E200338_ 652_ A.23.C20H14OSi2.90.4.set1d06_ 0-- 1292373 Valid 12-9-10 21:21:29 20-9-10 04:48:51 6.82 122.9 / 161.8 E200338_ 671_ A.23.C20H14S2Se.1.1.set1d06_ 1-- 1292373 Valid 12-9-10 21:21:29 18-9-10 16:21:47 6.94 125.5 / 158.4 E200338_ 798_ A.23.C20H14S2Si.23.0.set1d06_ 1-- 1292373 Valid 12-9-10 21:21:29 18-9-10 02:00:59 6.37 115.1 / 157.6 E200339_ 090_ A.23.C20H14SSi2.44.1.set1d06_ 0-- 1292373 Valid 12-9-10 21:21:10 16-9-10 04:48:56 6.29 113.6 / 145.9 E200339_ 070_ A.23.C20H14SSi2.38.set1d06_ 0-- 1292373 Valid 12-9-10 21:21:10 17-9-10 00:11:29 5.91 106.8 / 151.8 E200338_ 691_ A.23.C20H14S2Se.14.set1d06_ 0-- 1292373 Valid 12-9-10 21:21:10 17-9-10 18:09:37 6.92 125.1 / 159.5 E200339_ 045_ A.23.C20H14SSi2.24.2.set1d06_ 1-- 1292373 Valid 12-9-10 21:21:10 17-9-10 08:05:49 7.04 127.2 / 164.5 E200339_ 003_ A.23.C20H14SSi2.115.4.set1d06_ 1-- 1292373 Valid 12-9-10 21:20:46 15-9-10 18:34:14 8.27 149.5 / 190.9 E200338_ 773_ A.23.C20H14S2Si.19.0.set1d06_ 1-- 1292373 Valid 12-9-10 21:20:46 15-9-10 18:38:12 7.59 137.2 / 169.9 E200338_ 468_ A.23.C20H14OSi2.26.3.set1d06_ 1-- 1292373 Valid 12-9-10 21:20:46 15-9-10 18:34:14 6.59 119.0 / 146.1 E200337_ 909_ A.23.C20H14N2S.58.1.set1d06_ 1-- 1292373 Valid 12-9-10 21:20:46 15-9-10 18:34:31 6.74 121.7 / 151.5 An actual computed job within a task has a typical log sequence such as the below: [15:30:59] Starting job 14,CPU time has been restored to 12734.410000. [15:30:59] Starting new Job [15:30:59] Qink name = fldman [15:31:01] Qink name = gesman [15:31:02] Qink name = scfman [16:37:35] Qink name = anlman [16:42:48] End of Job [16:42:51] Finished Job #14 [16:42:51] Starting job 15,CPU time has been restored to 16789.830000.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Sep 25, 2010 10:28:25 AM] |
||
|
|
boulmontjj
Senior Cruncher France Joined: Nov 17, 2004 Post Count: 317 Status: Offline Project Badges:
|
Hi Sekerob
----------------------------------------Thanks for your reply. For points, it was just a question because it does not always seem logical. But i don't care about a few points less. In your reply, there is something i like a lot. Not in the answer itself but in your signature. I really like your HFCC banner. Is it something we can download from WCG site or is it your's and in that case, can we share it ? |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
H boulmontjj,
----------------------------------------The angry face smiley conveyed you were agitated, so hopefully that matter is cleared. The signature I borrowed from the HFCC website, news page. This research passed 30 million results last night.
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I don't know if it's your case, but I find that when HyperThreading is on, your granted credits are less than claimed, sometimes by a wide margin. So, if your PC is a Core i5/i7 (or even a Pentium 4) with HT on, that could explain the difference in points with your wingman.
|
||
|
|
boulmontjj
Senior Cruncher France Joined: Nov 17, 2004 Post Count: 317 Status: Offline Project Badges:
|
No, my pc that crubch CEP2 units run under Ubuntu and is a AMD 64X2 Dual Core 3800+ with no hyperthreading.
----------------------------------------I think soon, a quad 6600 or an other Core 2 Duo Intel will join it under Ubuntu. You're right Sekerob, i should have not use the angry face smiley but  because the last one that made me feel angry is now very old and the next one is may be not born yet.  Thanks for the banner. I have to say that i have not visited that site since a while.  |
||
|
|
martin64
Senior Cruncher Germany Joined: May 11, 2009 Post Count: 445 Status: Offline Project Badges: 
|
Sometime, something is strange with this research ; the points calculation. E200372_ 634_ A.24.C19H14N2OSSi.78.2.set1d06_ 1-- 619 Valide 22/09/10 01:14:07 25/09/10 04:59:45 5,54 103,0 / 168,8 E200372_ 634_ A.24.C19H14N2OSSi.78.2.set1d06_ 0-- 619 Valide 22/09/10 00:49:13 24/09/10 21:21:10 11,16 117,3 / 84,4 What does that mean ? What is the rule to calculate points ? It can be coincidence, but it is interesting that the #1 result has EXACTLY twice the number of points compared to #0 result. In a different thread there is a similar report. Maybe this is a bug in the credit granting mechanism? Regards, Martin  |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
No it's not! As noted, it's a proration method when one task did more work than the other.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
martin64
Senior Cruncher Germany Joined: May 11, 2009 Post Count: 445 Status: Offline Project Badges:
|
No it's not! As noted, it's a proration method when one task did more work than the other. Are you absolutely sure? Here is the example, see the 2 log files of the example mentioned above: There is one result that finishes normally after 16 tasks done. There is another one that finishes early due to the time limit reached. According to the expectations you set, the second one should get a couple of points LESS that the first one. But the opposite is the case: The job terminated due to reaching the time limits gets EXATLY TWICE the number of credits compared to the completed one. If there is any logic behind that, I need more explanation to understand.  Regards, Martin |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
hmmm, something is not kosher here. German Flu(ke), German (Bench)Mark inflation? On second look, smells of some slapping of tasks that are declared valid, with application of the invalid rule (half credit).
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
|