mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Aliquot Sequences (https://www.mersenneforum.org/forumdisplay.php?f=90)
-   -   Reserved for MF - Sequence 4788 (https://www.mersenneforum.org/showthread.php?t=11615)

smh 2009-04-28 14:11

[QUOTE=Andi47;171297]or run *only* the 25 digits level (i.e. 200 curves at B1=5e4), and THEN put it at the *end* of the queue.[/QUOTE]Preferably at the end of the 30 digit level queue after which it moves to the *end* of the 35 digit level queue, etc..

mdettweiler 2009-04-28 14:37

[quote=10metreh;171254]The t40 is still nowhere near completion because someone clicked "Stop" on it to make progress on another number.[/quote]
Sorry about that--I saw that apparently we'd already started a polynomial search on it, and figured that with all the B1=11M work that had been done, we had a full t40 complete which is plenty for this size number. Thus, I hit "stop" to avoid wasting the workers' time.

axn 2009-04-28 14:46

[QUOTE=Andi47;171297]or run *only* the 25 digits level (i.e. 200 curves at B1=5e4), and THEN put it at the *end* of the queue.[/QUOTE]
"ECM to high limits" already runs a number to 25-digit level (120 curves @ 9e4). So your suggestion is not very helpful (in fact, the whole "preparing" thing is useless once you've run "ECM to high limits").

I'd go with smh's suggestion, with an addition that smaller numbers should be run first before larger numbers.

henryzz 2009-04-28 15:31

[quote=axn;171310]"ECM to high limits" already runs a number to 25-digit level (120 curves @ 9e4). So your suggestion is not very helpful (in fact, the whole "preparing" thing is useless once you've run "ECM to high limits").

I'd go with smh's suggestion, with an addition that smaller numbers should be run first before larger numbers.[/quote]
people in general dont use "ECM to high limits" because it takes a long time because it uses only one core

10metreh 2009-04-28 15:35

What about inserting the new one at a random position in the queue? Then randomly rotate the queue every few minutes.

henryzz 2009-04-28 15:39

840 curves @ 1e6 with b2 9e8 = ~10% of t40
850 curves @ 3e6 with b2 7e9 = ~39% of t40
363 curves @ 11e6 with default b2 = ~53% of t40
we are now at ~102% of t40

jrk 2009-04-28 16:54

Here is my final poly (used msieve 1.41).

# norm 1.341618e-12 alpha -7.037308 e 7.077e-11

[code]n: 41044325296244095551421297177058739613282888480930079101779215677896599835732705957943543074751411525745638071656684388489293672281
skew: 501825.58
c0: 197813501589998296714291426614425
c1: 3594640378420732484907428415
c2: -3058781353740944451429
c3: -45948382794399547
c4: 18580727472
c5: 41760
Y0: -15794208298877591694042846
Y1: 271728935426531
rlim: 6000000
alim: 6000000
lpbr: 27
lpba: 27
mfbr: 54
mfba: 54
rlambda: 2.5
alambda: 2.5[/code]

Are the parameters I selected OK?

[QUOTE=mdettweiler;171249]However, if nobody's interested in going it alone, then of course a team sieve is still workable (though this time we'd have to be especially careful not to oversieve too much). I can volunteer for the postprocessing again if nobody else is particularly keen on doing it.[/QUOTE]
Let's do a team sieve, then.

schickel 2009-04-28 17:50

[QUOTE=henryzz;171331]840 curves @ 1e6 with b2 9e8 = ~10% of t40
850 curves @ 3e6 with b2 7e9 = ~39% of t40
363 curves @ 11e6 with default b2 = ~53% of t40
we are now at ~102% of t40[/QUOTE]You can add 200 curves @ 43e6 (default B2), so it's going to a nice split!

henryzz 2009-04-28 18:04

[quote=schickel;171362]You can add 200 curves @ 43e6 (default B2), so it's going to a nice split![/quote]
that on it's own is ~80% of t40

10metreh 2009-04-28 18:07

Meanwhile, I've started the sieve thread. I'm not a mod here [spoiler](yet)[/spoiler], so please could our worthy moderator put in the new poly and remove the "proposed" from its title?

jrk 2009-04-28 23:20

[QUOTE=schickel;171362]You can add 200 curves @ 43e6 (default B2),[/QUOTE]
Thanks


All times are UTC. The time now is 01:49.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2022, Jelsoft Enterprises Ltd.