mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > Cunningham Tables

Reply
 
Thread Tools
Old 2019-04-20, 03:14   #45
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

441810 Posts
Default

Quote:
Originally Posted by pinhodecarlos View Post
Could we have a step by step guide on how to install and then point the cores to the CADO server please? Much appreciated.

Carlos
Installing CADO for this job is one step removed from default "make":
Before compiling, open local.sh.example. Scroll all the way down, find the part that reads:
Code:
# For big factorizations, increase the size of variable.
# If you want to use large prime bound > 32, uncomment the following line
# FLAGS_SIZE="-D__SIZEOF_P_R_VALUES__=8"
delete the last # sign to uncomment that line. That sets CADO to allow LP bounds above 32.
Then save the file as local.sh (without the .example). Then, make.

To test your setup, run ./cado-nfs.py {input number}, of whatever size you feel like (100 digits is well under an hour on desktop quad-core, for example; 90 is just a few minutes). If you get factors, your setup works. If not, post to the CADO thread and hope some linux experts can help you out.

CADO by default runs on a whitelist setup, where I can enter the IP addresses of each client allowed to connect; that's how my jobs run right now. For this job, I think the best compromise of simplicity and security is to PM people the address of the server, rather than use the whitelist. I'm still reviewing documentation to see if there is a better way. The actual command to fetch work is merely invoking the cado-nfs-client.py script with the address and port number of the server. RichD and I have tested such connections on previous jobs, and it worked as advertised; recently I've discovered possibly a bit of a hiccup when the client is built with a different GCC than the server (but a workaround was provided on the CADO thread, just a flag on the command line that will also be shared when we're ready).
VBCurtis is offline   Reply With Quote
Old 2019-04-20, 03:49   #46
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

10001010000102 Posts
Default

A very brief test-sieve of Vebis' deg 6 poly with the 16f siever indicates something on the order of 60 core-years for the sieving; if 3 hours of test-sieve is accurate (it's not), that means we already have enough pledges to proceed with the factorization. Yield with 34/35LP looks fine on 16f, around 6 over Q=100-600M.

Vebis, could you share the P, admin, admax, nq, and root-opt level you used for the deg 6 poly select? It would be nice to not overlap search efforts.

I'm hoping Max has some spin to share soon; I think I have some time this weekend to try a CADO test-sieve (which for this rookie is just firing up the factorization, and hand-timing Q=1k workunits while noting the projected ETA after a few units are run). At the very least, I can learn what I=16 memory requirements are like, and hopefully also learn what lim's will keep RAM use at 13GB or so.
VBCurtis is offline   Reply With Quote
Old 2019-04-20, 06:29   #47
DukeBG
 
Mar 2018

3×43 Posts
Default

Quote:
Originally Posted by R.D. Silverman View Post
Yes. It was not suggested to run Cunninghams exclusively. However, I note
that as of this moment there are about ten (yes 10) different HC's running
simultaneously. How is this possible? Thus, when one number is near finishing
there are others still running full steam.....
My point is there are no different sets of HС numbers, AliquotSequences numbers, CullenWoodall numbers, etc. at yoyo that would pick up the slack in their set. The tabular display is just arbitrary. There's one set of all the numbers in ECM subproject and they already "pick up the slack" after one another.

As for the questions about why yoyo admin(s?) only added one TCP number and didn't add more yet, though it's clearly time for that, is something only they can answer, unfortunately. I can see why they wouldn't want to add 60 right away, but something like 5 at a time would be better.
DukeBG is offline   Reply With Quote
Old 2019-04-20, 07:11   #48
pinhodecarlos
 
pinhodecarlos's Avatar
 
"Carlos Pinho"
Oct 2011
Milton Keynes, UK

22·1,187 Posts
Default

Hi Curtis, thank you a bunch. I’ll play around oracle vm virtualbox next weekend. It’s been awhile since I’ve used Linux but I’ll call Paul Underwood to support me in case I need it. I suppose you can add me to your siever list, still not sure how many cores I can allocate since will know how much memory I can use after installing the VM. 16GB laptop, probably I’ll have to leave 2GB for Windows, I don’t know, I’ll need to investigate.

Last fiddled with by pinhodecarlos on 2019-04-20 at 07:21
pinhodecarlos is online now   Reply With Quote
Old 2019-04-20, 11:36   #49
R.D. Silverman
 
R.D. Silverman's Avatar
 
Nov 2003

723210 Posts
Default

Quote:
Originally Posted by DukeBG View Post
My point is there are no different sets of HС numbers, AliquotSequences numbers, CullenWoodall numbers, etc. at yoyo that would pick up the slack in their set. The tabular display is just arbitrary. There's one set of all the numbers in ECM subproject and they already "pick up the slack" after one another.
Therefore the probability of running a number at any given time from a particular subset
will be proportional to how many numbers that subset has queued......


Quote:
As for the questions about why yoyo admin(s?) only added one TCP number and didn't add more yet, though it's clearly time for that, is something only they can answer, unfortunately. I can see why they wouldn't want to add 60 right away, but something like 5 at a time would be better.
Indeed. 5 at a time would be terrific....
R.D. Silverman is offline   Reply With Quote
Old 2019-04-20, 14:10   #50
yoyo
 
yoyo's Avatar
 
Oct 2006
Berlin, Germany

11178 Posts
Default

Maybe I can bring some light in ecm hanling of yoyo@home.
In yoyo@home are 11 different ecm projects.Some of them have an empty queue. You also know tha runtime of a curve varies based on B1 and composite length.
If the number of ecm workunits goes down (below 1000) new ecm workunits are generated. My workunit generator tries to balance computing power equally between all ecm projects. So if new workunits must be generated the workunit generator checks which ecm project got least computing time in the last 5 days and taks curves from this project.

A cunningham number curve with B1=850M consumes much more computing time than a HC curve with B1=110M. Also composite length of HC is much smaller. This is the reason why HC has more runs in parallel.

But some times I do manual interactions and force a ecm projekt, e.g. during race times where I need longer running workunits.

yoyo

BTW: Up to B1=850M always 5 curves are in one BOINC workunit.
yoyo is offline   Reply With Quote
Old 2019-04-20, 14:22   #51
Max0526
 
"Max"
Jun 2016
Toronto

701 Posts
Default C207 poly -- 1.9

Quote:
Originally Posted by vebis View Post
Code:
n: 334377437706404684733884220732190564550147039308686365251932207592010868660163280944362994773042289837865752943086095333357861608005044927356022538834108845220872215597190097852981295940487497385774178731621
skew: 533132.085
c0: 13400871167573270650087884944996454257736480
c1: 170662016202584202246103429104502541792
c2: -106489302432833541662856860257198
c3: -1862164478135527803969537503
c4: 375011464350515249437
c5: 1161035722389120
c6: 848502000
Y0: -1694165626316187446160079002429297
Y1: 428100302665617359981
# MurphyE (Bf=3.436e+10,Bg=1.718e+10,area=5.469e+16) = 1.45e-08
# found by revision 9436c2ff8
# f(x) = 848502000*x^6+1161035722389120*x^5+375011464350515249437*x^4-1862164478135527803969537503*x^3-106489302432833541662856860257198*x^2+170662016202584202246103429104502541792*x+13400871167573270650087884944996454257736480
# g(x) = 428100302665617359981*x-1694165626316187446160079002429297
My best (deg-6)
Spin worked! This is a current record for C207 degree 6!
Code:
Y0: -1694165575355983617450319705011019
Y1: 428100302665617359981
c0: 11442175851420007438162901095682205599299200
c1: 486364944161428626833106069119830444060
c2: -717471225825159988151125727577076
c3: -1490458109841052560053038279
c4: 1246397994449168902237
c5: 1767059608845120
c6: 848502000
skew: 787799.64620
# lognorm 60.14, E 49.02, alpha -11.12 (proj -2.58), 2 real roots
# MurphyE(Bf=1.000e+07,Bg=5.000e+06,area=1.000e+16) = 1.93989983e-15
Max0526 is offline   Reply With Quote
Old 2019-04-20, 14:34   #52
vebis
 
vebis's Avatar
 
Oct 2015

22·17 Posts
Default

Quote:
Originally Posted by VBCurtis View Post
Vebis, could you share the P, admin, admax, nq, and root-opt level you used for the deg 6 poly select? It would be nice to not overlap search efforts.
I used the default c210.poly of the cado-nfs master branch ans let it run only for polyselect.

Last fiddled with by vebis on 2019-04-20 at 14:36
vebis is offline   Reply With Quote
Old 2019-04-20, 15:06   #53
Max0526
 
"Max"
Jun 2016
Toronto

2BD16 Posts
Default C207 file request

Quote:
Originally Posted by vebis View Post
I used the default c210.poly of the cado-nfs master branch ans let it run only for polyselect.
@vebis
Could you please send me a zipped file of everything that starts with c210 in your c207 work folder? (see image) Then I promise you to find a better poly in your data.
Attached Thumbnails
Click image for larger version

Name:	Screenshot from 2019-04-20 11-00-50.png
Views:	52
Size:	83.4 KB
ID:	20252  
Max0526 is offline   Reply With Quote
Old 2019-04-20, 15:55   #54
vebis
 
vebis's Avatar
 
Oct 2015

22×17 Posts
Default

Quote:
Originally Posted by Max0526 View Post
@vebis
Could you please send me a zipped file of everything that starts with c210 in your c207 work folder? (see image) Then I promise you to find a better poly in your data.

The whole thing is about 1.4G, i had to omit polyselect1 output.
Attached Files
File Type: tgz c207.tgz (570.7 KB, 41 views)

Last fiddled with by vebis on 2019-04-20 at 15:56
vebis is offline   Reply With Quote
Old 2019-04-20, 16:11   #55
RichD
 
RichD's Avatar
 
Sep 2008
Kansas

32×349 Posts
Default

My last partial run using a degree 5 only scored a 1.315. It looks like there is a good road to follow with degree 6 so I will stop my search.
RichD is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Coordination thread for redoing P-1 factoring ixfd64 Lone Mersenne Hunters 53 2020-10-30 19:39
big job planning henryzz Cunningham Tables 16 2010-08-07 05:08
Sieving reservations and coordination gd_barnes No Prime Left Behind 2 2008-02-16 03:28
Sieved files/sieving coordination gd_barnes Conjectures 'R Us 32 2008-01-22 03:09
Special Project Planning wblipp ElevenSmooth 2 2004-02-19 05:25

All times are UTC. The time now is 08:42.

Sat Oct 31 08:42:26 UTC 2020 up 51 days, 5:53, 2 users, load averages: 2.29, 1.89, 1.72

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, 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.