| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: FightAIDS@Home
Thread: FAAH WU Invalid.... |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 16
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I've noticed that one of my FAAH WU's is showing as invalid along with a greatly reduced granted credit - is someone able to let me know whether this is due to an actual error with the result, or is it being regarded for some reason as an outliner?
 The credit claimed is in line with all other WU's this machine crunches, and is using the standard 5.4.11 Boinc client, also the benchmark is the same as all my other Quad machines. The reason I say this is that the first 3 results in the list are claiming far less credit for some reason than the 2 showing as invalid (including mine). |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Currently, only genuinely invalid results are marked as such.
However, the fact that two from the same quorum were returned as invalid may be of interest to the techs. Also, the replication count at the top is wrong: the initial replication was 3, the other two were sent out later when the first 3 failed to validate. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks D, I was perhaps more worried that it was being regarded as an outliner, but this machine has proved very reliable and I don't like seeing invalids for any reason.
|
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Adywebb,
----------------------------------------Had a few FA@H jobs on the P4 that ran 3 times longer than any other in a set for unknown reason, but since the Outlier 50% penalty rule was discontinued, still got the mean of the others given there's now 'normalisation' (see member news). Only the real invalids get the 50% of the mean established by the valids in the quorum. For a while now 'Outliers' are not marked invalid i.e. the invalid got it's real meaning back. sekerob
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at May 6, 2007 5:36:23 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks Sek
 |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Ok - I have now had another FAAH result declared invalid under similar circumstances:
 Again my claim although normal, is substantially higher then the others in the quorum - in fact I find those low claims extremely odd for the amount of CPU time incurred  The reason why I am highlighting these is the fact that all my machines are near 100% error free - and these two invalids stand out because they are the ONLY two. The only thing I see in common between them is the fact that the other 'valid' results in the quorum are unusually low, which leads me to suggest that the only reason my result has been declared invalid is because my claim although normal for that machine, is totally out of line with the others. This seems to fly in the face of the fact that 'outliners' are supposedly no longer marked as invalid. Here are recent examples of that machines claims:  Please can one of the CA's bring this to the attention of theTechs and ask if they will look into this and come here and comment. Cheers, Ady  |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
I think it was mentioned in several threads that the benchmark in the 5.8 for the IOPS part went wobbly only reporting half in probably 4-5 out of 6 times. I've been testing 5.9.10 ALPHA !!!! on 2 machines and the Benchmark is now extremely stable and very little fluctuating due other things happening at time of the performance test, so the expectation for the next production release 5.10.x is optimistic.
----------------------------------------Following the sequence of transmission and return, it looks like the first 3 went inconclusive, invoked the sending of an extra copy. As that one had a 'No Reply', it caused a 5th copy to be send. After that was returned, yours was found REALLY invalid. That gives 50% of the mean. I see no fly in the face. So is my reading. If knreed would please review the set and advise.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 2 times, last edit by Sekerob at May 15, 2007 8:32:18 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I hear what you say Sek - but I find it strange that in the only 2 invalids I have, all the shown valid results have a very low credit claim
I have sifted through many many of my valid results and find none where all the valid claims by others in the quorum are so low. |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges: 
|
Adywebb,
Thank you for bringing this up - you just discovered a bug in the BOINC software. The reason you are invalid is because the rest of the results returned for that workunit are from Linux computer while your computer is Windows. When BOINC assigns a result to a computer. It first checks to see what platform has been assigned to the workunit and if it matches the computer. In this case, the workunit had not be assigned when it checked for you. However, at nearly the same instant the server was also checking for a Linux computer. It assigned the result to both of you at the same time and then update the workunit in the database first to windows and then almost immediately to Linux. Thus the remaining work was sent out to Linux. I will work on modifying this so that in the future this cannot happen again. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Glad my persistance in highlighting this has proved useful Kevin - I have to say it just stood out as being odd for some reason, but I just couldn't work out why!
Thanks for looking into this so promptly  Ady |
||
|
|
|