Secondpass Reaches 2M!!!

[vjs]

Just wanted to let everyone know that secondpass has now reached the 2M level.

Brings back memories for old SoB vets from years back...

When first pass was a 2M we were expecting another prime soon (this was long before the 5359•2^5054502+1 prime was found). :D During this period I was thinking there was a prime drought... We will see.

The good news there is alot less work compared to before, we have sieved alot deeper and eliminated 2k, not to mention we have faster clients and machines.

Happy hunting everyone, it will be interesting to see the error rate between 2M-3M.

[smh]

I haven't been following things lately, does this mean that there is a matching residue for all remaining numbers?

Or did testing just pass 2M, but several numbers without two matching residues need to bu run a 3rd time?

Are there any numbers on error rates?

[hhh]

The latter is the case, there are still some in the holepatch que, but no big deal. In a week it should be done or so.

I am waiting for the statistics, too. H.

[chris]

Greetings everybody,

I'm running holepatch for several weeks now. As of yesterday, about 1300+ tests were in queue, with 6-14 done each day.

Today there are 0 tests in residue-recovery. What happened?

Chris

Update: Client can't get a new block now that the current test is finished. Time for secondpass again :-)

[vjs]

I don't know, it would be funny if we actually found a prime :|party|: for

27653 I believe that was the only k being tested by holepatch.

Louie will probably say something, I would hope that the holepatch que be redirected to garbage or secondpass. I know I have 2 machines I can't access running that que.

[chris]

Well, if Louie is going to say something, it would be nice to comment the current Overall Min N 1028727 too.

Queue Name # Tests Min n Max n # Done n Increase
dropped-tests 5 9246404 13467751 58 +7198632
error-fix 435 5019120 7807623 0 n/a
first-pass 18301 9249341 9999967 131 +5990
garbage 1229 5021086 7594947 0 n/a
global 0 n/a n/a 0 n/a
largest-prime 13031 13467862 13999982 0 n/a
missing-test 0 n/a n/a 0 n/a
(not queued) n/a n/a n/a 4 n/a
residue-recovery 0 n/a n/a 10 +∞
second-pass 68616 2057712 4999946 115 +119170
Overall 107959 1028727 13999982 319 +-909815

(which, btw, is the same for a few month now i think)

Chris

[kfx]

If I am not mistaken, the first-pass queue have dropped a lot since yesterday, test for a given k may have been taken out of the queue.

[Frodo42]

A prime !!! :bouncy: :bouncy:

Oh well maybe I shouldn't be all that quick :blush:

[vjs]

Well the other problem is:

second-pass 68616 2057712 4999946 115 +119170

There are more than 68616 tests between 2.05M and 5M???

I think it's a server error, I'll e-mail louie.

[kfx]

Originally posted by vjs
Well the other problem is:

second-pass 68616 2057712 4999946 115 +119170

There are more than 68616 tests between 2.05M and 5M???

I think it's a server error, I'll e-mail louie.


Have you heard from Louie yet ?

[maddog1]

Well, all is revealed now :D
The clues for the upcoming prime for 27653 were here in this thread...some of you guys (chris, vjs, kfx, frodo42) detected the smoke, but couldn't really see the fire! But this time we came closest than ever before to the actual reality, even in the form of speculation :cheers:
LOL-Seems like Louie is doing a good job in hiding each prime find from us.

[kugano]

LOL-Seems like Louie is doing a good job in hiding each prime find from us.

Actually, we knew when we dequeued 27653 that people would figure it out. We could have "hidden" it a lot better, like we have in the past ;-) But now that we're down to so few multipliers, removing one of them gives us a major performance advantage, so we're thinking we should just dequeue immediately and let people sniff around until the announcement.

[vjs]

Yup, and it's probably best to do act as we did as users. Send an e-mail regarding the problem (Incase it wasn't prime) then continue on as normal. The worst thing that could possibly happen is we all start screaming prime and none is found.

Again Louie and Dave, I'd like to thank you both for not attemping to hide the discovery.

[chris]

Finally :D

vjs, have you heard from Louie about the 68K tests in secondpass? I expected to see a drop of 4000-5000 tests after the prime was confirmed, but this has not happened (yet?).

[kugano]

Again Louie and Dave, I'd like to thank you both for not attemping to hide the discovery. [/B]

And Mike. ;-)


vjs, have you heard from Louie about the 68K tests in secondpass? I expected to see a drop of 4000-5000 tests after the prime was confirmed, but this has not happened (yet?).

Shouldn't have expected that. We completely de-queued 27653 back on the 8th, including second-pass. Everything was removed in one shot. I'm not sure what the confusion's about.

[chris]

The 'confusion' on my side is that i have not seen secondpass drop the amount of tests i expected. 27653 had 6% of all assigned tests, so i expected (hoped?) for a somewhat 6 % reduction. I may simply have overlooked it, too, i mean, it seems i have somehow missed the de-que in secondpass. I noticed it in the other queues, however. So nevermind ;)

Good to hear everything is fine :thumbs:

[vjs]

I'll go through the numbers again but I'm pretty sure there was a drop in secondpass tests upon k elimination, that was how I verified my assumption that k=27653 was eliminated from que..

Use Mike_H's site for the number of tests between the bounds for the 9 remaining k... it looks correct currently. Actually there is less than there should be, this is due to some k/n pairs doublecheck above min secondpass, someone in the know could check further but here is my analysis.

Data from http://www.seventeenorbust.com/secret/

secondpass
67676 tests in que
Min n 2097020
Max n 4999946


Now calculation from Mikes site... http://www.aooq73.dsl.pipex.com/scores_p.htm

Number of secondpass test remaining between doublecheck and 5M

75746 for 10K if you remove the number of tests between 2M<n<5M for k=27653 you get... roughly 70258 tests remaining (not quite correct probably closer to ~70K b/c I didn't consider those 2M<n<secondpass for k=27652).

Even if this number is ~70K

That basically means that sure k=27653 was removes and we have already doublechecked ~2500 tests between secondpass and 5M (4.5%).





_____________________________________

Side note this should be enough tests to calculate a decent, error rate for secondpass to 5M ...