k=33661 blocks starting up?

[Halon50]

My dual MP-2200+ just grabbed a k=33661 block. Are we starting up work for the new k value already?

:cheers:

[jjjjL]

right now there are 11 new blocks out just to see how things go.

there are still about 950 tests left for k=27653. you can see about what n values those tests correspond to based on the new graph i added to the stats page.

i was assigning k=33661 blocks for about 2 hours today but the priority has been given back to the k=27653 blocks for now. you just happened to be lucky enough to get one ahead of the pack ;)

also - as noted on the homepage, your cEM rate will fall with the new k=33661 until the n values get higher. that is normal so don't drop the block just because the cEMs decreased... that should happen.

-Louie

[Halon50]

This is true, I'm getting about 40% of the cEM rate on the new k value compared to the old.

Thanks for the reply!

[Firebirth]

Ok... this is really just for vizualization of the graph, but couldn't the number of remaining numbers have a red colored graph on top of teh blue one for numbers that have been sent out to the clients; in that way, it would be more clear, when we would be getting the 33661 blocks!

Of course if the whole coloring thing is a bad idea (I suggested it before by coloring the 33661 in a new color in the old statistics!); this is too a bad idea.:jester:

Tarje

[Mystwalker]

I don't know if we need a graph for that.
Basically, the values sent out to the clients are the lowest ones still available. So you just have to look up "Blocks currently pending" and look at the graph. Maybe a horizontal line should to the trick.

I would prefer a table that shows how much n values are missing - divided in 100,000 intervals or so. It *is* possible to get that info out of the graph, but I think it's hard to tell the correct count.
Maybe this can be integrated in the big-k.shtml - page. Just a suggestion - fixing the Linux client ought to be of higher priority. But if you are fed up with that... :p

[priwo]

I think i would prefer Mystwalkers idea.

priwo

[Halon50]

Just thought I should bring this up again: I had to install a new HD on a machine that previously used a failing IBM Deskstar drive. It's now running on a new Maxtor while I wait for a (second) warranty replacement.

The new install just picked up a number for k=33661, so I assume this means we're underway! Are there any plans on how to display the two k values concurrently on the stats page?

[Mystwalker]

Well, ~600 values are still missing and currently 603 blocks are pending, so I think we started the new k. :|party|:

I can't wait till one of my clients get a new one. ;)

[smh]

Yes, the server is handing out tests for the new K
I've got one. (to be honest, i stoped a test at 99,9% on one pc (only 1 hour) and waited till the last N's were assigned before i completed the test)

Unfortunately it's neither updated in the graph and the stats.

[nuutti]

smh: What is your n and how long one test takes ?

It is interesting to see how long it will take when we are
n=>1,000,000 (day or two?) and when we are n=>2,000,000.


Yours,

Nuutti

[smh]

Originally posted by nuutti
smh: What is your n and how long one test takes ?

It is interesting to see how long it will take when we are
n=>1,000,000 (day or two?) and when we are n=>2,000,000.


Hmm, dunno, 650K something IIRC. The PC is at home.
But i think there are others who have already completed some N's of the new K. Maybe they can post part of their log file?

[Tax Bracket]

I got one today.

I'll post the log when it's finished (which it nearly is already...)

[Halon50]

It's taking about 2.5 hours for a single n value (1 block and change) around the 650k mark, on a Duron-850.

[Tax Bracket]

[Tue Nov 19 12:50:27 2002] got proth test from server (k=33661, n=657360)
[Tue Nov 19 16:03:42 2002] logging into server
[Tue Nov 19 16:03:42 2002] login successful
[Tue Nov 19 16:03:42 2002] n.high = 578540 . 1 blocks left in test
[Tue Nov 19 16:29:44 2002] residue: B5D8D752E748472E
[Tue Nov 19 16:29:44 2002] completed proth test(k=33661, n=657360): result 3

[nuutti]

Thanks.
I got my first one :
[Tue Nov 19 19:34:36 2002] got proth test from server (k=33661, n=681288)

estimated time to complete about 4 hours (with my old poor PIII 450)

Yours,

Nuutti

[Mystwalker]

Got one, too - at my PC @ work. ATM it's a tad more than one block to complete. KcEMs are quite low, though - as promised...

[Bellbox]

Here's a few so you can get an idea... I'm averaging about 100,000 cEMs/sec less than on the smaller blocks.


[Tue Nov 19 13:52:48 2002] got proth test from server (k=33661, n=690624)
[Tue Nov 19 14:57:29 2002] logging into server
[Tue Nov 19 14:57:29 2002] login successful
[Tue Nov 19 14:57:30 2002] n.high = 524151 . 1 blocks left in test
[Tue Nov 19 15:18:01 2002] residue: B5FE7B161C557C47
[Tue Nov 19 15:18:01 2002] completed proth test(k=33661, n=690624): result 3
[Tue Nov 19 15:18:01 2002] connecting to server
[Tue Nov 19 15:18:01 2002] logging into server
[Tue Nov 19 15:18:03 2002] requesting a block
[Tue Nov 19 15:18:04 2002] got proth test from server (k=33661, n=702240)
[Tue Nov 19 16:19:24 2002] logging into server
[Tue Nov 19 16:19:24 2002] login successful
[Tue Nov 19 16:19:24 2002] n.high = 506954 . 1 blocks left in test
[Tue Nov 19 16:43:00 2002] residue: D0D48D0B344E45E8
[Tue Nov 19 16:43:00 2002] completed proth test(k=33661, n=702240): result 3
[Tue Nov 19 16:43:00 2002] connecting to server
[Tue Nov 19 16:43:01 2002] logging into server
[Tue Nov 19 16:43:01 2002] requesting a block
[Tue Nov 19 16:43:05 2002] got proth test from server (k=33661, n=709200)
[Tue Nov 19 17:40:38 2002] logging into server
[Tue Nov 19 17:40:38 2002] login successful
[Tue Nov 19 17:40:39 2002] n.high = 497052 . 1 blocks left in test
[Tue Nov 19 18:05:13 2002] residue: D9A4039187DBCBBE
[Tue Nov 19 18:05:13 2002] completed proth test(k=33661, n=709200): result 3
[Tue Nov 19 18:05:13 2002] connecting to server
[Tue Nov 19 18:05:13 2002] logging into server
[Tue Nov 19 18:05:13 2002] requesting a block
[Tue Nov 19 18:05:15 2002] got proth test from server (k=33661, n=716640)
[Tue Nov 19 19:02:04 2002] logging into server
[Tue Nov 19 19:02:04 2002] login successful
[Tue Nov 19 19:02:05 2002] n.high = 486785 . 1 blocks left in test
[Tue Nov 19 19:28:56 2002] residue: D7BD18C9AAD5A5DC
[Tue Nov 19 19:28:56 2002] completed proth test(k=33661, n=716640): result 3
[Tue Nov 19 19:28:56 2002] connecting to server
[Tue Nov 19 19:28:56 2002] logging into server
[Tue Nov 19 19:28:57 2002] requesting a block
[Tue Nov 19 19:28:58 2002] got proth test from server (k=33661, n=724776)
[Tue Nov 19 20:24:31 2002] logging into server
[Tue Nov 19 20:24:32 2002] login successful
[Tue Nov 19 20:24:32 2002] n.high = 475918 . 1 blocks left in test
[Tue Nov 19 20:54:17 2002] residue: 8BA8727746F9F8BD
[Tue Nov 19 20:54:17 2002] completed proth test(k=33661, n=724776): result 3

[Halon50]

The main stats page is now showing somewhat anomalous reports:


Mathematical Statistics (as of Nov 20 03:18 pm EST)
--------------------------------------------------------------------------------

Current multiplier under test 27653
Largest n inclusively tested 2172560
Largest n tested 2999769
Largest n in progress 2999517
Blocks currently pending 478 blocks
% exps that failed trial factoring 0.263%
Proth tests completed 7812 tests
The "Largest n tested" is actually higher than the "Largest n in progress"! I guess this is expected though, as the fast machines move on to the new k blocks, and slower ones finish up work on the current k.

I know you guys are probably approaching or in the middle of finals right now - or at least in the "crunch" week where all the final projects are due. Any word on when stats will switch or include the new k value?

At any rate, take your time, get through the "Real Life" stuff, and we'll patiently await your next developments! :cheers:

[nuutti]

this is also from main stats page so someone is modifying pages right now?
k=33661 is tested up to n=649319
k=27653 is tested up to n=2172560


also
Blocks currently pending 477 blocks
peaked at 606 so I guess that (606-477)=129 machines have switched to new k.

Yours,

Nuutti

[smh]

There is a graph now for the new K.

It would be even better if the "Mathematical Statistics " sections could be extended to reflect the new K.

I guess the 475 blocks that are currentely pending is only for K=27653 (maybe current blocks would be a better naming.

I guess it will take a while before all of those 475 will be tested. the first couple of hundreds will be done within a couple of days, but there always seem to be some exponents that keep being assigned to the wrong people.

Today, after testing a K=33661 number i got a K=27653 number twice. I moved one to a no net machine to speed things up a little.

[jjjjL]

yeah i just modified the stats a bit.

there are not full stats for 33661 because i'm trying to come up with a better way to display stats for all 17 values at once. probably a slightly more complex table but in the long run will have more data.

i have to change the stats backend a bit for it to work well. right now it's a complete hack how it's doing it.

also, you may notice i just added the k-graph for 33661.

the k graphs now only show pending blocks while before it showed even unassigned blocks. showing all blocks makes the 33661 graph pretty uninteresting.


you may notice that a few more values have been "reserved" for SB over at http://www.prothsearch.net/sierp.html we expect to be getting even more soon so there is a high likelihood that we will just begin testing basically all of them.

in a future client, there will probably be a way to choose which value you'd like to work on if you like.

before that though, expect there to be a client with a new block format. it will make caching work across all OSes and make it possible to have multiple blocks out at once. basically there will be no reg keys and all data for config will be in a .ini or .conf file and all each block will be it's own file.

-Louie

[Halon50]

Woo hoo! Looking great! :cheers:

Also, taking off the period from the Prothsearch.net URL above makes it work. :rolleyes:

[nuutti]

>I guess it will take a while before all of those 475 will be tested.
>the first couple of hundreds will be done within a couple of days,
>but there always seem to be some exponents that keep being
>assigned to the wrong people.

Heh. Milestone guy ?
I think that situation is not bad, because we have only two
weeks expiry time.
Improvment suggestion : if n<=1,000,000 then expiry period could be only one week or even less.

Yours,

Nuutti