View Full Version : New First-Pass tests
jjjjL
03-01-2009, 11:07 AM
It's time we start testing brand new numbers. Double-check is pretty solid now.
New numbers are in the queues and will check out as you finish your current assignments (unless you configured your clients to purposely pull only double-checks.)
Cheers,
Louie
It's time we start testing brand new numbers. Double-check is pretty solid now.
New numbers are in the queues and will check out as you finish your current assignments (unless you configured your clients to purposely pull only double-checks.)
Cheers,
Louie
Seems to run... except that for 21181, max n is left behind at n=15947012 while all others are around 16,008 K (and in the worktodo files there are some k=21181 in the interval 16,000-16,008 K).
jjjjL
03-03-2009, 02:01 AM
Seems to run... except that for 21181, max n is left behind at n=15947012 while all others are around 16,008 K (and in the worktodo files there are some k=21181 in the interval 16,000-16,008 K).
Yeah, I'm aware of that. I had to do something special before I queued k=21181. It's in there now.
Cheers,
Louie
MikeH
03-03-2009, 11:54 AM
Louie, great to see you back.
Out of interest, is the information on http://www.seventeenorbust.com/secret/ correct, specifically for the first-pass queue? It claims the bottom of the first-pass queue is 13258604, and with 66600 tests in the queue, there are about 6000 tests before we get back to n=16M. Yet I read that folks are getting n=16M+ tests.
Also the next.txt file (which I use for the sieve stats) gives the same value as the secret page.
Could you take a look please.
Cheers,
Mike.
jjjjL
03-03-2009, 10:26 PM
Louie, great to see you back.
Out of interest, is the information on http://www.seventeenorbust.com/secret/ correct, specifically for the first-pass queue? It claims the bottom of the first-pass queue is 13258604, and with 66600 tests in the queue, there are about 6000 tests before we get back to n=16M. Yet I read that folks are getting n=16M+ tests.
Also the next.txt file (which I use for the sieve stats) gives the same value as the secret page.
Could you take a look please.
Cheers,
Mike.
Hey Mike,
Yes, surprisingly the numbers are right. There are a few first-pass tests that truly are first-pass down near n=13M... but the density of tests between 13.2M - 14.xM is pretty low... will probably collapse back to previous n-levels in a couple weeks.
And the 16M+ tests are all going to the new server I'm testing with the GIMPS client, not the larger SB pool.
Cheers,
Louie
MikeH
03-04-2009, 05:37 AM
Now I understand. :thumbs:
Cheers,
Mike.
bigmanwhs
03-10-2009, 09:42 AM
Are new the tests being pulled only from first-pass right now? The stat page looks like second-pass is still at the same priority. I expected to be in the new n-range soon.
Death
03-10-2009, 10:08 AM
oh can i guess? sieve scoring will raise up as current n window move on?
jjjjL
03-10-2009, 01:14 PM
Are new the tests being pulled only from first-pass right now? The stat page looks like second-pass is still at the same priority. I expected to be in the new n-range soon.
Ah yes, you are correct. The server was assigning from both queues but mostly second-pass. It now only assigns second-pass to those who request it and first-pass is the default, not both.
That was an important update that needed to be made. Thank you for the helpful feedback!
Cheers,
Louie
shauge
03-30-2009, 01:44 PM
Louie, great to see you back.
Out of interest, is the information on http://www.seventeenorbust.com/secret/ correct, specifically for the first-pass queue? It claims the bottom of the first-pass queue is 13258604, and with 66600 tests in the queue, there are about 6000 tests before we get back to n=16M. Yet I read that folks are getting n=16M+ tests.
Also the next.txt file (which I use for the sieve stats) gives the same value as the secret page.
Could you take a look please.
Cheers,
Mike.
Hey Mike,
Yes, surprisingly the numbers are right. There are a few first-pass tests that truly are first-pass down near n=13M... but the density of tests between 13.2M - 14.xM is pretty low... will probably collapse back to previous n-levels in a couple weeks.
And the 16M+ tests are all going to the new server I'm testing with the GIMPS client, not the larger SB pool.
Cheers,
Louie
Have there been a change of plans? Now there have been almost 4 weeks more with no progress in the in the sieve stats, because the current next.txt is far below where it was before the FP queue was moved back. Last time FP was done, back in November 2008, I believe, it stopped at 16M.
I am currently factoring over 4000 tests between 14.5M and 16M and there are much more FP PRP tests than that to do before 16M. AFAIK, it is only the legacy application that does these tests now and emails have been sent out encouraging people to move to the new app, Prime95. So, it unless something is changed, it will take months until the next.txt catches up.
Meanwhile, the Prime95 apps are continuing at full speed from the previous FP stop point and are now at approx. 16132294, ref. http://www.seventeenorbust.com/stats/rangeStatsEx.mhtml.
MikeH, is there a possibility that you manually can enter the real "next.txt" value based on rangeStatsEx? E.g. once a week or so?
jjjjL
03-30-2009, 03:46 PM
Why do you want next.txt to be updated? I can make it use the higher number if that would be better. The density of tests below 16M is very low and almost not worth considering for any sort of stats.
Cheers,
Louie
shauge
03-30-2009, 04:59 PM
Why do you want next.txt to be updated? I can make it use the higher number if that would be better. The density of tests below 16M is very low and almost not worth considering for any sort of stats.
Cheers,
Louie
MikeH has made what I think is one of the greatest DC-Stats ever. I just wonder what he needs to get it going again, since there is no next.txt for the queue used by the new server for the GIMPS client :)
chris
04-03-2009, 01:36 PM
How are these stats possible?
Queue Name # Tests Min n Max n # Done n Increase
dropped-tests 1727 14633671 16142780 45 +2263737
error-fix 0 n/a n/a 0 n/a
first-pass 62745 14633671 19999951 65 +21779
wolfemancs
04-03-2009, 01:39 PM
Different k?
Just guessing.
MikeH
04-03-2009, 06:30 PM
MikeH has made what I think is one of the greatest DC-Stats ever. I just wonder what he needs to get it going again, since there is no next.txt for the queue used by the new server for the GIMPS client
Many thanks, you are too kind. I aim to please. :thumbs:
At the moment I'm manually updating with the max tested n on the project stats page, couple of times per week.
I hoping the queues will get merged and it'll all go back to being automatic again some time in the not too distant future.
shauge
04-04-2009, 04:49 AM
Many thanks, you are too kind. I aim to please. :thumbs:
At the moment I'm manually updating with the max tested n on the project stats page, couple of times per week.
I hoping the queues will get merged and it'll all go back to being automatic again some time in the not too distant future.
Thanks MikeH, your effort is appreciated :thumbs:
I also hope the queues soon will merge :)
Powered by vBulletin® Version 4.2.4 Copyright © 2025 vBulletin Solutions, Inc. All rights reserved.