| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: The Clean Energy Project Forum
Thread: Invalid CEP WUs twice as large as valid ones yet get same or fewer points? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 7
|
|
| Author |
|
|
AppleTV
Cruncher Joined: Dec 7, 2008 Post Count: 45 Status: Offline |
E000059_ 950A_ 000d4000m_ 2-- Valid 12/29/08 02:04:53 12/30/08 07:52:17 14.51 253.6 / 206.3
E000059_ 950A_ 000d4000m_ 0-- Valid 12/26/08 05:48:49 12/27/08 21:21:34 14.67 185.5 / 206.3 E000059_ 950A_ 000d4000m_ 1-- Invalid 12/26/08 05:48:04 12/29/08 01:07:20 30.45 253.3 / 206.3 The longer WU packet (1--) took 30.45 hours yet got the same credit as a WU packet that took less than half that (14.xx hours for the other two)... Is this because WUs that are deemed invalid only get half credit? |
||
|
|
Dark Angel
Veteran Cruncher Australia Joined: Nov 11, 2005 Post Count: 728 Status: Offline Project Badges: 
|
As I understand it, invalid work gets half credit, yes.
----------------------------------------Only outright errors get nothing.  Currently being moderated under false pretences |
||
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3716 Status: Offline Project Badges: 
|
In fact there are two different things which puzzle you at the same time:
----------------------------------------- an invalid WU has been granted the same credit as the valid ones while, normally, it should have been credited only half (103.15) - two WUs have claimed the same credits although one has run twice longer than the other. The latter is perfectly normal. The faster client had a benchmark twice bigger than the slower, which is logical and what you see shows that the computation of claimed credits is fair, at least for this particular case. The objective being to reward clients for the work they have done we are here in the ideal case where two very different machines are claiming the same number of credits for the same task, which is perfect (it is not always so perfect, unfortunately!). About the former, it is possible that the techs have tweaked the CEP validator to "pay" invalid results the same as the valid ones until they are able to decrease the number of abnormal results for which the clients are not responsible. That will be less work for them when/if they start giving back all the credits that these various bugs have forbidden members to get. Cheers. Jean. |
||
|
|
AppleTV
Cruncher Joined: Dec 7, 2008 Post Count: 45 Status: Offline |
In fact there are two different things which puzzle you at the same time: - an invalid WU has been granted the same credit as the valid ones while, normally, it should have been credited only half (103.15) - two WUs have claimed the same credits although one has run twice longer than the other. The latter is perfectly normal. The faster client had a benchmark twice bigger than the slower, which is logical and what you see shows that the computation of claimed credits is fair, at least for this particular case. .. well, in this case, all three WUs were from the same client which operates with dual single-core processors (hence the _0 and _2 units taking nearly equal times to complete).. Why the _1 unit took twice as long is the oddity here, and perhaps the reason why the result is invalid?? |
||
|
|
bieberj
Senior Cruncher United States Joined: Dec 2, 2004 Post Count: 406 Status: Offline Project Badges: 
|
AppleTV,
There is no way that the same client has done the same job three times. Each job is done by different computers. One of the jobs were done on a computer that ran twice as fast as the other one. The invalid was done on the third computer that ran into a problem. The third computer could have a bit flip that occurred by chance or something is wrong with its memory or disk space or the task that got downloaded didn't download cleanly and the bad download that was not detected. Or something else. Or perhaps an undetected bug in the CEP code itself. I realize that I am speculating here. Everything is working as it is designed to work. The invalid result is likely to be analyzed to determine that this a software bug and if so, will have the bug fixed. Just don't worry about it. JB |
||
|
|
AppleTV
Cruncher Joined: Dec 7, 2008 Post Count: 45 Status: Offline |
There is no way that the same client has done the same job three times. OK, it was a misunderstanding on my part that when I clicked on the invalid unit, I thought it was showing me three parts of the same job generated from my machine, instead I now understand it's three replications of the same unit with the invalid result from my machine tossed in for comparative purposes.. Now it makes sense to me why the result is invalid! |
||
|
|
bieberj
Senior Cruncher United States Joined: Dec 2, 2004 Post Count: 406 Status: Offline Project Badges:
|
Glad to help clear up misunderstandings.
Crunch away...  |
||
|
|
|