If you can't get a block

[vjs]

Not sure whats going on but if you can't start a new test it's possible that the firstpass que has not been repopulated.

Alternatives

Run secondpass for a while, to do this change your username to the following

usernameQQQsecondpass

In my case this would be vjsQQQsecondpass




The other alternative is to enter the username garbage or secret

[Alien88]

We're going to run on second-pass for a little while..

[Vato]

I'd appreciate an update on how long? and/or why? And I suspect I'm not alone :-)

Not meaning to be shirty or anything - I'd just rather be finding a prime rather than double checking......

[Frodo42]

I agree that it would be nice with an update on this.

Just to make sure no one misunderstands this ... the reason we do secondpass at all is that there is a chance of finding a prime that was missed because on an earlier error.

I seem to remember some back of the envelope rough estimates that said that around this point there would be approximately the same chance of finding a prime in secondpass and in firstpass, over the same given time.
Anyhow we need more data on the error-rate than has been released to say anything about this for sure.

Anyhow even though it at the max-n may look like a set-back I doubth that it really is for the project as a whole.

Of course some stat-maniacs will be sad to see that they score less with the lower n-tests ... but that is the same for all right now.

[vjs]

Yup, from that back of the envelope calculation at n=~8M we were suppose to have secondpass around 2.6-4.5M depending on what you believed. Now that we are at n=10M secondpass should be much higher than it is.

It shouldn't take us long to do all of secondpass to the n=4M level anyways (not sure how long we are going to do secondapss). But I think bringing the level up to where is should be as an absolute minimum.

BTW there are roughly 8K tests before we reach 3M with secondpass. And 24215
tests between 3M<n<4M, considering we are now doing roughly 500 tests per day and that rate is increasing, it wouldn't take us long to reach 4M.

Also bringing the secondpass level up helps other aspects of the project as well.

- gives people more time to sieve and factor.

_____________________________________

One more possibility, what if we actually found a prime co-incidental with the n=10m running out? Say a few days before, it would make sence to switch everyone to secondpass as the prime is verified.

[Alien88]

The main reason we're running second-pass for a little bit is because we never intended for it to get as far behind as it is. Now is just a decent time to get caught up on it..

as for how long? probably not too long.. i can't give a 'real' estimate as to when we'll switch back over as of yet.

[Frodo42]

>Alien88

I am not sure if the "workaround way" of population of the dropped test que (http://www.free-dc.org/forum/showthread.php?s=&threadid=10116) necesarry or has the server now been "correctly" configured to also deal out second-pass test to people without the QQQsecondpass in their nick?

[Joh14vers6]

Originally posted by Frodo42
>Alien88

I am not sure if the "workaround way" of population of the dropped test que (http://www.free-dc.org/forum/showthread.php?s=&threadid=10116) necesarry or has the server now been "correctly" configured to also deal out second-pass test to people without the QQQsecondpass in their nick?
It was nescessary when the global and dropped-tests queues went empty, but configuring the seccondpass-queue to be accessible when both queues are empty is a good idea.

[Alien88]

Originally posted by Frodo42
>Alien88

I am not sure if the "workaround way" of population of the dropped test que (http://www.free-dc.org/forum/showthread.php?s=&threadid=10116) necesarry or has the server now been "correctly" configured to also deal out second-pass test to people without the QQQsecondpass in their nick?

Sorry for not being clear when I originally said it. When I said we were going to do second pass for a while - we reconfigured the server to hand out second-pass instead of first pass..

so yes, people will get 2nd pass without using QQQsecondpass.

[Frodo42]

Originally posted by Alien88
so yes, people will get 2nd pass without using QQQsecondpass.
Thank you so much for clarifying this :thumbs:

[Nuri]

May be we should announce this at the news section.

http://www.seventeenorbust.com/news/

Some people might think this is not a temporary issue and leave the project. Informing them that the second pass tests will be assigned by the server for a limited time only, and giving an ETA if possible, even if something vogue like we are planning to switch back to normal course within the next couple of weeks, might keep them crunching.

[vjs]

And also a remark regarding speed points.

"The scoring system for small n tests will be scaled in a future update, reported client rates do not acurrately reflect client contribution."

[maddog1]

Got to agree with Nuri and VJS, this has to make the news!
Otherwise, all the new users and stats freaks are bound to run away, don't expect everyone to bother to read through the forum...

As for me, since I was one of the few people running secondpass before it was enforced, nothing changes for me ;)

[Haranaka]

Originally posted by Frodo42

Of course some stat-maniacs will be sad to see that they score less with the lower n-tests ... but that is the same for all right now.
Well I'm one of those stats-maniacs and I don't agree with you on this point. Ofcource it's true that everyone is going slower now that we're secondpassing, but the problem lies in the fact that the difference in overall score becomes more importent because everybodies output drops.

A small example with some numbers:
Team X has an total of 1000 points, and a daily output of 10 points.
Team Y has a total of 700 points and a daily score of 20 points.

At this rate team Y will overtake X in 30 days

But what happens when you change both their daily output by 50%? It takes team Y now 60 days to clear the gap between X and Y. That is my point.

[Greenbank]

The dropped-tests queue now have a minimum n of 7M so the difference in scores will be much less now.

[Keroberts1]

also when the new scoring version comes out it will be evening out al of the differances so that equal CPU time earns equal credit. Don't worry all will be evened out in time.

[Haranaka]

Originally posted by Keroberts1
also when the new scoring version comes out it will be evening out al of the differances so that equal CPU time earns equal credit. Don't worry all will be evened out in time.
New scoring version? Can someone give me some more details about that? If it's only CPU time that counts I'll guess I bring my old PII 266 online to crunch...

[hhh]

PII? Did you consider sieving? There you should be finding something from time to time. Or head over to the PrimeSierpinskiProject, (sorry everybody, but it has been founded to fit in old machines). It's at www.mersenneforum.org.

As for your question, it has been discussed already, in a few words: Until now, firstpass tests scored more then Secondpass. Now, everybody is doing SP. One day, the scoring-injustice will be corrected, making climb in the stats those who ran SP before. (But in fact just correcting their discrimination).

My english is kind of crappy today, can't explain it better. You can look up the forum, if you want details.

Yours H.

[Keroberts1]

There will ofcourse always be the fact that unless a computer completes a test (reguardless of how many blocks were completed) then no credit will be given to the user. While working on a test a user may see his credit but if he cancells the test with out finishing it he/she will not recieve any credit for the work. Tests must be completed to help the project and ifyou're not helpingthe project then you don't deserve stats. I'm just saying this because a PII 266 would require almost a year to finish a single test meanwhile if it was prime there would have been many many tests wasted while waiting. If yo uwanted ot run it on secondpass though that would both benefit the project and earn you stats credit that would eventually be equivalent to the ammount of power contributed, plus just maybe you could find a prime.(i believe chances are actually better of finding one in second pass right now than in regulars testing.) I would recommend using the (username)QQQsecondpass option because then the regular server starts handing out first pass tests again you wouldn't want your slow machines getting any tests they weren't likely to finish.

[stim]

Originally posted by Haranaka
New scoring version? Can someone give me some more details about that? If it's only CPU time that counts I'll guess I bring my old PII 266 online to crunch...

No, the new scoring method will not just be cpu time. I'm not sure exactly what it will count on, but from memory from what I've read in the forum, it still used the k and n numbers but does it is a more fair way than the current method. I think all that Keroberts1 meant with the cpu time comment was that if the same processor works for 2 days on second pass tests, it will get the same credit as working on first pass for 2 days (not the way it works at the moment as we are all experiencing). Of course, two different speed processors wont get the same credit though.

So you can leave your old PII in the cupboard :)

[Keroberts1]

i decided that since ihave a stable system agtai ni was going to try a little double check PRPing but when i started the client i recieved test 27653 7790289 is this a double check test? Did it come from the dropped test que? Where did this come from? Hoping one of the admins would be able to answer this. I've already gotten a couple blocks done so I'm just going ot let it finish but if it is a ouble check I would like ot cancel it and get a double chec ktest that is with the rest of the DC testers.

OK on second thought we already have a prime for this test so I'm abandoning it now Why are these tests still being handed out?

[engracio]

I hope the scoring adjustment happens soon. My secondpass test seems like I'm going backward or losing ground even though I was doing almost the same amount of wu. Someday will happen, I just know it.


e:)

[DigitalConcepts]

Originally posted by Keroberts1
OK on second thought we already have a prime for this test so I'm abandoning it now Why are these tests still being handed out?
Good catch Keroberts1. The answer should be interesting.

[Nuri]

I've made a table based on stats and graphs in the project web site.

Please note that First Pass vs Second Pass distribution is a rough estimate based on the graphs. ;)

As you can see, there are not many pending first pass tests for primes found. I guess they still exist simply because they were and still are pending since the time the prime was found. And as a wild guess on why such a test was handed out nowadays, I guess it might simply be because it was one of those tests and it was later dropped either because of the 90 day limit or simply flushed by previous user.


The table also provides (if my reasoning is right) a glimpse of the current situation in terms of the number of first pass tests and second pass tests that are currently pending.

[vjs]

Yup, It's difficult to both remove the prime tests from the server and still give people credit for them. Either solution is an evil. But we can be certain that no tests for primes are being handed out.

The last prime was certainly better than the previous ones, Louie stopped handing those out as soon as one was discovered.

[Keroberts1]

well i recieved a test for a prime yesterday so it is very certain that they are still being handed out either because the yare going into the dropped test que or being placed in the error fix que either way people shuld not still be getting those tests... Its just a waste of work...