stephenbrooks.orgForumMuon1GeneralNew Member with quarantined results?
Username: Password:
Search site:
Subscribe to thread via RSS
Freewolf
2004-01-20 20:05:17
What is a Quarantined result?  A new member of TA had this on his first upload The client queued the result and says "running quarantined result 4 of 5 times.
Freewolf
2004-01-20 21:50:34
Never mind, I found out why Big Grin
Diabolos.de
2004-01-21 14:25:38
I say cheating.  If its all OK i say Im sorry.  But i cant understand that in the last time, the results "running" up so high.  Still today my best result is 9,83xxx, and muon runs ever and ever and ever...
Earlier i had rechecked results and since some weeks muon never do that.  So why all other (US-Distibuted) make high results and rechecked results.

Signature: No Thx iam not ill.
[OCAU] badger
2004-01-21 21:59:34
definately cheating Wink
I just foudn that i had submitted a good enough result to get me up to the #7 spot in transfer (for a short while anyway!), but I barely get rechecks anymore.

It is all down to chance, we are now pretty close to the maximum transfer for this setup, and so it takes a "lucky" mutation to get a result higher than the best in your results.dat, and with the new version, most people have a results.dat which has the latest highest score in it.

www.BadgerMotorsport.tk Proudly sponsored by GRX-Computers
Fritz
2004-01-24 08:43:11
Folks,

The secret to our rapid rise in the DPAD stats is that we are running a great number of fast systems on the program.  As near as we can estimate we have spent about $250,000 team wide on upgrades over the last year because we are serious about supporting worthwhile Distributed Computing programs.  If thats cheating, then we are guilty. 

Best Regards,

Fritz
US-Distributed
edco
2004-01-24 09:08:41
Hey Fritz,

Well said!...

I'll contact Stephen and get his take on these accusations from Diabolos.de I hope this his personal opinion and not the view of the Rechenkraft team.

Sincerely,

edco
US-Distributed

Who are these guys?...
Diabolos.de
2004-01-24 15:36:24
Ok, i wrote "its all OK i say im sorry".
Im Sorry.  But you must understand me.  Earlier my results are in the Top10 - Top20. Since some weeks, muon runs and runs and my results going better but not so quick and high like other guys.  I had think, the results gone to quick to high.  I also saw that more and more crunchers from US-Distributed where in the Top10, since some weeks.  And so i had think that cannot be right.

Tell us now please, what is the best size for the results.dat and what can we do to optimizing, to get also good results like US-Distributed.

In German we say: Asche auf mein Haupt.

Signature: No Thx iam not ill.
Stephen Brooks
2004-01-24 16:15:29
Ash on your head???  Big Grin Razz

I was surprised too when the graph (which had been flat for weeks) suddenly started stepping up again.  It's the sort of thing that would happen if the queueing mechanism was broken and results with larger statistical variation got submitted, but I checked the database and all appears to be fine.  Another explanation is that these results come from a different lineage to the previous best ones and this breed has only just become developed enough to push through the previous maximum.  Someone here made a spreadsheet for comparing the similarity of results so maybe it would be possible to find out if there are two such different groups in the current sample100.

Anyway, it may be that US-Distributed are pooling results and have come across a better alternative.  You can try resetting your results.dat (or taking it away to another folder) and starting just from the project sample100 because that will contain a lot of the higher sort of result.

HB Pencils, also sold as "Moron's Choice" Graphite Cigars.
Freewolf
2004-01-24 18:21:57
BTW, neither I nor the member I was asking about are members of US-Distributed.  We belong to Team Anandtech; we don't cheat in any project and do not put up with it.

[This message was edited by Stephen Brooks on 2004-Jan-25 at 3:32.]
Fritz
2004-01-24 20:04:29
Stephen,

As far as I know, we are not intentionally pooling out results.  Some of us do have intermittent internet connections, but there is no intentional sharing. 
We may have just stumbled on something new, and because so many of us are running for the rally it looks like sharing.

Regards,

Fritz
Herb[Romulus2]
2004-01-25 03:34:25
The spreadsheet is here http://webwi.de/data/queue.xls (1M, it's maintained daily.

Some things have changed and it's far to complex to explain it here. 
As a general recommendation: all results which have a grey cell in the last column shall form a sample results.dat , this would give a maximum of different variations from all current breeds without significant repetitions.  These sample selections vary only little since 2 weeks and the current top breed is dominating now the top 250 rechecked results.  But there are still some interesting deviations around.

Hey Diabolis, I had the same effect, I'd been month always in the top 20 and now nothing goes, just bad luck, keep it easy.  Wink

-------------------------------
I'd say more, but I can't reach the keyboard from the floor.
Stephen Brooks
2004-01-25 09:22:52
Well, I wasn't saying it was a bad thing to share results, just that that would explain why one particular team appeared to be ahead.  Actually, looking at the stats, while there are some US-Distributed members in the top 10 by yield, there are a lot of others too.  So no need to be paranoid, I guess.

HB Pencils, also sold as "Moron's Choice" Graphite Cigars.
amdxborg
2004-01-25 13:08:08
Yeah man, Freewolf said it [TA] isn't moving up by cheating, we're moving up by crunching like mad and by being a great TeAm...who will be nr.1!!!  Cool

______________________________
AMD Athlon XP2600+, Soltek SL-75FRN2-L, Corsair 512MB XMS PC3500

[TA]amd.borg
Proud member of [TA] SETI@home & DPAD
: contact : - - -
E-mail: sbstrudel characterstephenbrooks.orgTwitter: stephenjbrooksMastodon: strudel charactersjbstrudel charactermstdn.io RSS feed

Site has had 25161367 accesses.