mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > Factoring

Reply
 
Thread Tools
Old 2014-11-13, 14:31   #1
fivemack
(loop (#_fork))
 
fivemack's Avatar
 
Feb 2006
Cambridge, England

3·19·113 Posts
Default M991 reservations thread

34-bit large primes are not necessary, and I have pretty high confidence that Q values greater than 2^30 aren't necessary either; we'll be happy with a billion relations, and I'm getting yields of at least 1.8 rel/Q from

Code:
n: 265093018953022478194238260373068717560582078760717338146236227613636234616964158720771659078210842628613505064050422312061555900485118292436714714643935928917834010049633442636550112497704144055980572214129219808682422285559933440692755932893569335559067600176369
skew: 1
c6: 2
c0: -1
Y1: 1
Y0: 46768052394588893382517914646921056628989841375232
lpbr: 33
lpba: 33
mfbr: 96
mfba: 66
alambda: 2.6
rlambda: 3.6
alim: 400000000
rlim: 400000000
using gnfs-lasieve4I16e -r. Plan on one modern-core month per million Q. The choice of skew affects only sieving speed, but Batalov is trying to find the best value; this calculation is big enough that 1% would be really useful.

Please make no more reservations after 23/3 and get everything uploaded by 27/3

I'll start by reserving 200M to 248M, which I expect to take about a month; I will start sieving in two weeks when I've finished Fib(1373).

Sieving binaries:
Example run command:
Code:
start /low /min gnfs-lasieve4I16e.exe -o 248-249.out -r M991.poly -f 248000000 -c 1000000 -v
Please make reservations adjacent to current reservations; ideally let's finish 50M-500M before going outside it.

Code:
Gimarel           85- 90 13/02/2015   05/03/15  15843215u
Gimarel           90- 96 22/01/2015   13/02/15  18786592u
VBCurtis          96- 98 15/01/2015   [done]
VBCurtis          98-100 06/01/2015   [done]
balamber         100-140 06/01/2015   01/02/15 112804053u
Gimarel          140-150 18/12/2014   13/02/15  27849881u
balamber         150-199 06/12/2014   13/01/15 124821546u
VBCurtis         199-200 14/11/2014   15/01/15 ?
fivemack         200-248 13/11/2014   22/02/15 114111349u
legendarymudkip  248-252 13/11/2014  [half done; fivemack finishing]
RichD            252-260 13/11/2014   08/12/14  18770126u
dubslow          260-263 14/11/2014   31/12/14   7033760u
Gimarel          263-268 16/11/2014   01/12/14  11606078
Batalov          268-280 13/11/2014   17/11/14  27661457
kracker          280-284 13/11/2014  [done by 17/01/2015]
YuL              284-296 25/11/2014   22/01/15  27063665
RichD            296-300 19/11/2014   14/12/14   8952231u
balamber         300-324 15/11/2014   06/12/14  52968941
Gimarel          324-330 01/12/2014   21/12/14  12999205u
RichD            330-342 12/12/2014   14/01/15  25669870u
dubslow          342-346 31/12/2014   16/01/15   8530508u
RichD            346-350 14/01/2015   02/02/15   8489722u
dubslow          350-352 23/01/2015   09/02/15   4230134u
RichD            352-356 24/01/2015   06/02/15   8449304u
YuL              356-376 26/01/2015   20/03/15  41616819
dubslow          376-378 26/01/2015   03/02/15   4141710u
RichD            378-382 02/02/2015   23/02/15   8246331u
fivemack         382-394 06/02/2015   03/03/15  24385140u
RichD            394-396 08/02/2015   18/02/15   4071858u
dubslow          396-400 12/02/2015   07/03/15   8111523u
balamber         400-401 06/12/2014   06/12/14   2023335
balamber         401-410 29/01/2015   08/02/15   8034120
fivemack         410-422 22/02/2015   10/03/15  23702448u
fivemack         422-438 25/02/2015
RichD            438-440 26/02/2015   12/03/15   3907508u
fivemack         440-450 07/03/2015   23/03/15   19365333u
balamber         500-501 06/12/2014   06/12/14   1856430
balamber         600-606 06/12/2014   10/12/14  10366733

Last fiddled with by fivemack on 2015-03-24 at 12:06 Reason: update results/reservations as of 12/3
fivemack is offline   Reply With Quote
Old 2014-11-13, 16:39   #2
chris2be8
 
chris2be8's Avatar
 
Sep 2009

2,179 Posts
Default

Is "skew: 1" optimal? The default would be about 0.89. Have you tested it?

Chris
chris2be8 is offline   Reply With Quote
Old 2014-11-13, 17:47   #3
Batalov
 
Batalov's Avatar
 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2

2×4,787 Posts
Default

I'll take my usual regions of interest (around powers of 2, where bugs happen).
I'll take 268-280M for starters.
Batalov is offline   Reply With Quote
Old 2014-11-13, 17:58   #4
Gimarel
 
Apr 2010

18510 Posts
Default

Quote:
Originally Posted by chris2be8 View Post
Is "skew: 1" optimal? The default would be about 0.89. Have you tested it?
The best skew is 1.17 according to msieve.
Gimarel is offline   Reply With Quote
Old 2014-11-13, 18:39   #5
Batalov
 
Batalov's Avatar
 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2

100101011001102 Posts
Default

The conventional skew=|c0/c[deg]|^(1/deg) formula does exactly give us 0.89.

Maximizing E_value is also not much better at learning the optimal skew. But it is very fast and scientifically sound (within the assumptions of a certain nature of the sieving region). The top at the optimal value is extremely flat; a take home lesson is that if you started a batch with some, even if slightly suboptimal skew, don't kill it (just run it through as planned; optionally change the next batch); the difference will be miniscule (e.g. for 1<= skew <= 1.3; 0.89 is however more perceptibly suboptimal):
Code:
skew 1.17, size 1.459e-14, alpha 2.482, combined = 4.453e-15 rroots = 2
skew 1.15, size 1.459e-14, alpha 2.482, combined = 4.453e-15 rroots = 2
skew 1.10, size 1.459e-14, alpha 2.482, combined = 4.452e-15 rroots = 2
skew 1.20, size 1.459e-14, alpha 2.482, combined = 4.452e-15 rroots = 2
skew 1.25, size 1.459e-14, alpha 2.482, combined = 4.451e-15 rroots = 2
skew 1.30, size 1.459e-14, alpha 2.482, combined = 4.450e-15 rroots = 2
skew 1.00, size 1.459e-14, alpha 2.482, combined = 4.446e-15 rroots = 2
skew 1.45, size 1.459e-14, alpha 2.482, combined = 4.441e-15 rroots = 2
skew 0.89, size 1.459e-14, alpha 2.482, combined = 4.433e-15 rroots = 2
The reason why the optimal skew is different from the initial estimate (which is 0.89) is totally different. One reason is that 3LP is used. The other is that the sieving region will be different from the assumptions of the default calculation. Only trial sieving in many points and statistical treatment (because the data is very noisy and will be easily misleading) would give a truly meaningful answer. I am trial sieving now, Tom probably has done already.

Last fiddled with by Batalov on 2014-11-13 at 19:24 Reason: (some trivial E sampling)
Batalov is offline   Reply With Quote
Old 2014-11-13, 19:11   #6
pinhodecarlos
 
pinhodecarlos's Avatar
 
"Carlos Pinho"
Oct 2011
Milton Keynes, UK

2×3×829 Posts
Default

This post might be useful to run several instances of the linux client.

http://www.mersenneforum.org/showpos...7&postcount=89

Last fiddled with by pinhodecarlos on 2014-11-13 at 19:12
pinhodecarlos is offline   Reply With Quote
Old 2014-11-13, 19:48   #7
legendarymudkip
 
legendarymudkip's Avatar
 
Jun 2014

7816 Posts
Default

I'll take 248M-252M
legendarymudkip is offline   Reply With Quote
Old 2014-11-13, 20:56   #8
RichD
 
RichD's Avatar
 
Sep 2008
Kansas

3,449 Posts
Default

I'll take 252M-260M.
RichD is offline   Reply With Quote
Old 2014-11-13, 23:44   #9
kracker
 
kracker's Avatar
 
"Mr. Meeseeks"
Jan 2012
California, USA

32×241 Posts
Default

I'll take 280-284 for now. How much memory does it need in total?
kracker is offline   Reply With Quote
Old 2014-11-14, 02:04   #10
Dubslow
Basketry That Evening!
 
Dubslow's Avatar
 
"Bunslow the Bold"
Jun 2011
40<A<43 -89<O<-88

722110 Posts
Default

So we're using the old lasieve4e binaries, no 5 or f or anything?
Dubslow is offline   Reply With Quote
Old 2014-11-14, 07:04   #11
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

10011100101002 Posts
Default

Per the first post and its enlarged type, correct- standard 16e siever.
VBCurtis is online now   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
M991 press release? fivemack Factoring 9 2015-07-04 20:12
M991 discussion jasonp Factoring 85 2014-12-01 20:01
M2376 reservations thread fivemack ElevenSmooth 53 2008-04-28 08:53
Range reservations thread MooooMoo Twin Prime Search 38 2006-06-08 04:12
Deutscher Thread (german thread) TauCeti NFSNET Discussion 0 2003-12-11 22:12

All times are UTC. The time now is 23:22.


Wed Oct 27 23:22:37 UTC 2021 up 96 days, 17:51, 0 users, load averages: 1.11, 0.92, 0.94

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.