| Author |
Message |
|---|
|
|
|
What happened to these work units (Sophie Germain)
http://www.primegrid.com/workunit.php?wuid=95561113
I have another 26 identical to that |
|
|
RytisVolunteer moderator
Project administrator
Send message
Joined: 22 Jun 05
Posts: 2639
ID: 1
Credit: 21,321,147
RAC: 66
                    
|
|
That's a "feature" of BOINC that apparently hides WU information for apps that are using adaptive replication so that it is harder to cheat. I'll see what can be done about that.
____________
|
|
|
|
|
|
Yeah I have about 15 waiting with all Suppressed Completion in them. sucks if they are prime :(
____________
John M. Johnson "Novex" |
|
|
 HAmsty Volunteer tester
Send message
Joined: 26 Dec 08
Posts: 132
ID: 33421
Credit: 12,510,712
RAC: 0
          
|
|
any news on this feature? i have over 520 SGS Wus pending :(
____________
 |
|
|
valtercVolunteer tester
Send message
Joined: 30 May 07
Posts: 119
ID: 8810
Credit: 4,380,731,669
RAC: 3,582,916
                   
|
any news on this feature? i have over 520 SGS Wus pending :(
Still the same behaviour.... That's the only LLR subproject doing this way. Is there any particular reson for this?
|
|
|
RytisVolunteer moderator
Project administrator
Send message
Joined: 22 Jun 05
Posts: 2639
ID: 1
Credit: 21,321,147
RAC: 66
|
|
BOINC is hiding the data about WUs from this project because we're only doing a single pass on some of them. If it was known that a WU you're doing is only having a single pass it would be easy to simply cheat and return bogus data, hence BOINC is hiding it until it is validated.
____________
|
|
|
|
|
BOINC is hiding the data about WUs from this project because we're only doing a single pass on some of them. If it was known that a WU you're doing is only having a single pass it would be easy to simply cheat and return bogus data, hence BOINC is hiding it until it is validated.
which is completely silly since one can see his claimed credit on his own list of pending results as long as they are pending.
and of course this is in no way a measure to prevent users from cheating - it will only not be visible to others.
so in fact this "feature" will probably be considered nice by those who are doing this.
another brilliant idea... |
|
|
samuel7Volunteer tester
Send message
Joined: 1 May 09
Posts: 89
ID: 39425
Credit: 257,416,435
RAC: 0
                
|
BOINC is hiding the data about WUs from this project because we're only doing a single pass on some of them. If it was known that a WU you're doing is only having a single pass it would be easy to simply cheat and return bogus data, hence BOINC is hiding it until it is validated.
which is completely silly since one can see his claimed credit on his own list of pending results as long as they are pending.
and of course this is in no way a measure to prevent users from cheating - it will only not be visible to others.
so in fact this "feature" will probably be considered nice by those who are doing this.
another brilliant idea...
The point is to hide the number of replications from all users until validated. If you can see it on your pending list you can hardly cheat with that task anymore.
____________
 |
|
|
|
|
The point is to hide the number of replications from all users until validated. If you can see it on your pending list you can hardly cheat with that task anymore.
yup - one would need to take the next one...
... of course this is ends up about the same as waiting for a pending one to get validated and see what you have squeezed out.
just my 5 cents: fixed credits for a certain amout of work done successfully is the only way. for sure this requires a validator that's really working.
lately i was able to watch results being reported with totally erratic runtimes simply because this whole concept is really fouled up now - walltime 30 minutes, reported 6, and granted for those 6 according to that really crazy benchmurks. :(
|
|
|
samuel7Volunteer tester
Send message
Joined: 1 May 09
Posts: 89
ID: 39425
Credit: 257,416,435
RAC: 0
|
The point is to hide the number of replications from all users until validated. If you can see it on your pending list you can hardly cheat with that task anymore.
yup - one would need to take the next one...
... of course this is ends up about the same as waiting for a pending one to get validated and see what you have squeezed out.
You'd have to get lucky the first time as you won't be the only checker for a while once caught cheating. More info in the Trac wiki
I agree that fixed credits are the best way, but I understand that it's probably tough for this project to implement in LLRs with changing n values. However, the current benchmarks should be replaced by something more reliable.
____________
|
|
|
|
|
I agree that fixed credits are the best way, but I understand that it's probably tough for this project to implement in LLRs with changing n values
really?
go ask those guys running projects with precalculated credits on WU's with differing difficulty - SIMAP and QMC come to my mind here..
|
|
|
