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)

Mini-Geek 2009-07-21 14:45

[quote=bsquared;182075]But does reporting work affect the workers behavior?[/quote]
Yes. Once the DB sees that enough work has been reported, it takes the number out of the VHL queue. I don't see any problem with what just happened. 2000 curves at B1=43e6 should eliminate or greatly reduce the need for ~1250 more curves at B1=3e6 (what the DB would have run had it continued).

bsquared 2009-07-21 15:03

[quote=Mini-Geek;182078]Yes. Once the DB sees that enough work has been reported, it takes the number out of the VHL queue. I don't see any problem with what just happened. 2000 curves at B1=43e6 should eliminate or greatly reduce the need for ~1250 more curves at B1=3e6 (what the DB would have run had it continued).[/quote]

Huh, didn't know that.... learn something every day! Thanks.

I can make up for those lost DB curves tonight with another couple thousand at 43M.

10metreh 2009-07-21 15:20

[quote=bsquared;182082]I can make up for those lost DB curves tonight with another couple thousand at 43M.[/quote]

Me thinks we should start poly searching after those are done.

axn 2009-07-21 15:51

[QUOTE=10metreh;182085]Me thinks we should start poly searching after those are done.[/QUOTE]

I suggest ~8000 curves at 43M. Poly search can be started even now, but sieving should wait till the full t50 ECM.

Greebley 2009-07-21 15:56

My 15 attempts at 110M all failed. Not that this is significant compared to the rest but I thought I would report it anyway.

Mini-Geek 2009-07-21 16:09

[quote=axn;182090]Poly search can be started even now,[/quote]
Wouldn't it be more efficient to finish all ECM, then do all poly search, then all sieving, then all filtering, instead of letting them overlap?

Andi47 2009-07-21 16:28

[QUOTE=Mini-Geek;182096]Wouldn't it be more efficient to finish all ECM, then do all poly search, then all sieving, then all filtering, instead of letting them overlap?[/QUOTE]

Yes - as it is still possible that ECM finds a factor, and in that case any work done with GNFS would be wasted.

axn 2009-07-21 16:51

[QUOTE=Mini-Geek;182096]Wouldn't it be more efficient to finish all ECM, then do all poly search, then all sieving, then all filtering, instead of letting them overlap?[/QUOTE]

Unless you want to finish it as early as possible (calendar time).

My proposal was intended to minimize the wastage, while trying to speedup the process.

Mini-Geek 2009-07-21 19:28

[quote=axn;182106]Unless you want to finish it as early as possible (calendar time).

My proposal was intended to minimize the wastage, while trying to speedup the process.[/quote]
Hm, makes sense. I suppose it's the best way with what we've got.
However, if someone were to make a program to automate and communicate between all the CPUs working on this c157, it could keep ~100% efficiency and best possible wall time. Imagine each CPU checking in every 1-5 curves. Once the server determines that enough ECM has been run, it switches all the CPUs over to poly searching as soon as they report their batch of curves finished. Use similarly small batches and switchover method for poly searching and sieving, (with automated reporting of found polys and relations) then stop all workers once the relations gathered reaches how many should be needed and run the filtering, then if necessary resume sieving on all workers.
I don't suppose anything like that exists, does it? I don't suppose any of our resident coders would put forward the time for an app like that? :whistle:
Hey, I can dream can't I?

Batalov 2009-07-21 20:53

Isn't it called "ECMNET server"? :smile:

jrk 2009-07-21 21:35

Look!
[code]Using B1=43000000, B2=388112953420, polynomial Dickson(30), sigma=492002632
Step 1 took 169903ms
Step 2 took 85269ms
********** Factor found in step 2: 270389991140767419113595201012871830378223721738111
Found probable prime factor of 51 digits: 270389991140767419113595201012871830378223721738111
Probable prime cofactor 13097232778890996127721661850226147250346721968629381919532862503065919780081186230385499514937725380715199 has 107 digits
[/code]

mdettweiler 2009-07-21 21:36

[quote=Batalov;182142]Isn't it called "ECMNET server"? :smile:[/quote]
Except that ECMNet doesn't do polynomial searches or sieving.

bsquared 2009-07-21 21:50

[quote=jrk;182148]Look!
[code]Using B1=43000000, B2=388112953420, polynomial Dickson(30), sigma=492002632
Step 1 took 169903ms
Step 2 took 85269ms
********** Factor found in step 2: 270389991140767419113595201012871830378223721738111
Found probable prime factor of 51 digits: 270389991140767419113595201012871830378223721738111
Probable prime cofactor 13097232778890996127721661850226147250346721968629381919532862503065919780081186230385499514937725380715199 has 107 digits
[/code][/quote]

:w00t:

Very nice!

axn 2009-07-21 21:54

[QUOTE=bsquared;182151]:w00t:

Very nice![/QUOTE]

Indeed. But another c155 :sad: Did someone say 8000 curves @ 43M?

Mini-Geek 2009-07-21 21:56

[quote=Batalov;182142]Isn't it called "ECMNET server"? :smile:[/quote]
Unless I'm mistaken, that only runs ECM, not all the sections with smart switchover.

Batalov 2009-07-21 22:16

Well, if we had an ECMNET server running, by now all willing ECMers would have thrown quite a few curves on the c155, not wasting time on the solved c157! The server (with a manager) would have coordinated that.

Of course it doesn't do poly sel, but it doesn't prevent us from setting it up. imho.

fivemack 2009-07-21 22:23

running 600@43e6 should be done in about 24 hours

jrk 2009-07-21 22:23

[QUOTE=jrk;182148]
[code]Using B1=43000000, B2=388112953420, polynomial Dickson(30), sigma=492002632
Step 1 took 169903ms
Step 2 took 85269ms
********** Factor found in step 2: 270389991140767419113595201012871830378223721738111
Found probable prime factor of 51 digits: 270389991140767419113595201012871830378223721738111
Probable prime cofactor 13097232778890996127721661850226147250346721968629381919532862503065919780081186230385499514937725380715199 has 107 digits
[/code][/QUOTE]

I computed the group order of the factorization:
[code][2 4]

[3 1]

[19 1]

[2473 1]

[17317 1]

[339331 1]

[608297 1]

[859459 1]

[1621679 1]

[4828303 1]

[4983967657 1]
[/code]

So the factor could have been found with as little B1=5e6, B2=5e9.

Mini-Geek 2009-07-22 00:51

[quote=jrk;182156]I computed the group order of the factorization:
...[/quote]
How is this computed from the sigma? (or otherwise computed)
Here are some links related to this:
[URL]http://factordb.com/search.php?id=61049323[/URL] (the c157)
[URL]http://factordb.com/search.php?id=66234919[/URL] (the p51)
[URL]http://factordb.com/search.php?id=66581132[/URL] (the c51 you denoted by the factorization given)
[URL]http://factordb.com/search.php?id=66584546[/URL] (a p25 produced by c51-p51)
[quote=jrk;182156]So the factor could have been found with as little B1=5e6, B2=5e9.[/quote]
How much ECM was run at lengths closer to this? That would've been quite something to have found a p51 factor with B1 as low as 5e6! (which is between the B1 values for 40 and 45 digits)
[URL="http://factordb.com/search.php?id=66584546"] [/URL]

jrk 2009-07-22 01:11

[QUOTE=Mini-Geek;182162]How is this computed from the sigma? (or otherwise computed)[/QUOTE]
See: [url=http://www.mersenneforum.org/showthread.php?p=84923]this thread[/url].

Greebley 2009-07-22 01:35

But isn't every elliptic curve is different so a different sigma would be a different factorization because each elliptic curve is a different size group? Thats the way I understood it.

The real point is there was a fifty digit number so you needed that size (which we chose) to find it (and we did). At 5e6 very few curves would find it.

jrk 2009-07-22 02:05

[QUOTE=Greebley;182167]But isn't every elliptic curve is different so a different sigma would be a different factorization because each elliptic curve is a different size group? Thats the way I understood it.

The real point is there was a fifty digit number so you needed that size (which we chose) to find it (and we did). At 5e6 very few curves would find it.[/QUOTE]

That's right.

bsquared 2009-07-22 03:15

[CODE]
Run 76 out of 250:
Using B1=11000000, B2=35133391030, polynomial Dickson(12), sigma=2937282173
Step 1 took 50670ms
Step 2 took 19890ms
********** Factor found in step 2: 24991157461902436638364677738700981780530893
Found probable prime factor of 44 digits: 24991157461902436638364677738700981780530893
Probable prime cofactor 770133409328510593934098647054112348863648709974876203302121061034340674324233272425893132240154136634104602673 has 111 digits
[/CODE]

Partway through 4000 curves at 11M, this popped out.

jrk 2009-07-22 03:17

Good one.

Batalov 2009-07-22 03:17

a c134 so far

axn 2009-07-22 04:20

Who's this SB that is submitting all these curves??

Andi_HB 2009-07-22 04:28

Maybe SB = Serge Batalov?

Batalov 2009-07-22 04:54

correct

10metreh 2009-07-22 06:44

We've still got that 2^5 * 3 :rant:

And we've reached 160 digits :rant:

I think we've done enough ECM from what's in the database; does anyone want to crack the C134 themselves or shall we make it a team GNFS?

jrk 2009-07-23 01:26

[QUOTE=10metreh;182192]I think we've done enough ECM from what's in the database; does anyone want to crack the C134 themselves or shall we make it a team GNFS?[/QUOTE]

No reply. I guess it will be a team sieve, then.

mdettweiler 2009-07-23 17:07

Hi all,

I am pleased to announce that the nplb-gb1.no-ip.org FTP server is back online again. It should be all ready for the next team sieve. :smile:

Max :smile:

10metreh 2009-07-23 17:15

[quote=mdettweiler;182370]Hi all,

I am pleased to announce that the nplb-gb1.no-ip.org FTP server is back online again. It should be all ready for the next team sieve. :smile:

Max :smile:[/quote]

Would you like to do the poly search for the c134?

Andi_HB 2009-07-23 18:53

The best poly from Msieve till now for the c134

R0: -74345499857629306588651703
R1: 526918406537017
A0: 1476115900135771763787951266874992
A1: 626824231219973690300198460
A2: -13207605174359180727636
A3: -12643447556511725
A4: 5363656602
A5: 20400
skew 702726.68, size 6.658692e-013, alpha -6.594295, combined = 4.814539e-011

mdettweiler 2009-07-23 20:46

[quote=10metreh;182373]Would you like to do the poly search for the c134?[/quote]
No thanks--I'm afraid my resources are quite tied up at the moment. Right now, going through the available subproject sequences one at a time is all I can handle.

axn 2009-07-23 22:00

[QUOTE=Andi_HB;182409]The best poly from Msieve till now for the c134

R0: -74345499857629306588651703
R1: 526918406537017
A0: 1476115900135771763787951266874992
A1: 626824231219973690300198460
A2: -13207605174359180727636
A3: -12643447556511725
A4: 5363656602
A5: 20400
skew 702726.68, size 6.658692e-013, alpha -6.594295, combined = 4.814539e-011[/QUOTE]

The best I got from pol51 was a 4.37. I think this one is good enough.

jrk 2009-07-24 00:23

I searched 770-775K with msieve and got a 3.734e-11.

def-nm-params.txt says a "good" poly for c134 is 3.48e-11. So a 4.814e-11 poly must be great. We should start sieving with it.

Suggest these parameters? I did not optimize them:

[code]n: 46334491956043710225938330967034053252570375961099723291677005874807967585788533290212203831083417490921313409294721505021600000357391
Y0: -74345499857629306588651703
Y1: 526918406537017
c0: 1476115900135771763787951266874992
c1: 626824231219973690300198460
c2: -13207605174359180727636
c3: -12643447556511725
c4: 5363656602
c5: 20400
skew: 702726.68
rlim: 10000000
alim: 10000000
lpbr: 28
lpba: 28
mfbr: 56
mfba: 56
rlambda: 2.5
alambda: 2.5
[/code]

axn 2009-07-24 01:03

Not so fast! We've got competition.
[CODE]BEGIN POLY #skewness 171420.77 norm 2.86e+018 alpha -6.73 Murphy_E 4.91e-011
X5 519120
X4 -149547362004
X3 22880757084328834
X2 3084767389898150786204
X1 -728499066352347968303249289
X0 -48696807949270208376362448497955
Y1 702872416394437
Y0 -38915962909433205107082626
M 42072142647191243626145050524826061496202553462804199881021111232312303878027902907455010638887626416613882212096551514178349273986581
END POLY

BEGIN POLY #skewness 160469.13 norm 1.80e+018 alpha -6.45 Murphy_E 4.70e-011
X5 519120
X4 -76686274404
X3 10179548269110898
X2 4419407981607320670062
X1 -512845919649890281305746985
X0 -26649405539109601030995687817650
Y1 702872416394437
Y0 -38915943179101604498841599
M 42072142647191243626145050524826061496202553462804199881021111232312303878027902907455010638887626416613882212096551514178349273958510
END POLY

BEGIN POLY #skewness 164341.85 norm 1.87e+018 alpha -6.47 Murphy_E 4.70e-011
X5 519120
X4 -80312327604
X3 10618202363121250
X2 4375833371795537802728
X1 -525133153734057584913310499
X0 -34014987672382883340184185370965
Y1 702872416394437
Y0 -38915944161014370201870088
M 42072142647191243626145050524826061496202553462804199881021111232312303878027902907455010638887626416613882212096551514178349273959907
END POLY
[/CODE]

10metreh 2009-07-24 07:24

It's only a C134, so sieving time will be so small (compared to a C157) that I think we should just take the best: anyone want to test the parameters for the 4.91?

mdettweiler 2009-07-24 11:25

FYI: The FTP server will be available, as before, for this team sieve. I've got the directory "c134-relations" all set.

bsquared 2009-07-24 13:43

[quote=10metreh;182492]It's only a C134, so sieving time will be so small (compared to a C157) that I think we should just take the best: anyone want to test the parameters for the 4.91?[/quote]


Over a tiny test interval, the msieve poly was quite a bit better:
[CODE]
4.814 poly
total yield: 5863, q=5001001 (0.02095 sec/rel)

4.91 poly
total yield: 4305, q=5001001 (0.02157 sec/rel)
[/CODE]

I think the parameters look good, I suggest we go with the poly file jrk posted. 5 MQ should be more than enough for this one. I'll reserve 2M-3M.

jrk 2009-07-24 16:58

[QUOTE=bsquared;182525]I suggest we go with the poly file jrk posted.[/QUOTE]
The poly was actually found by Andi_HB (see post #325), I only added the parameters.

jrk 2009-07-28 07:29

[QUOTE=jrk;183104]Factors:

[code]prp54 factor: 129616193467788152689435837807707789229325708626916411
prp81 factor: 357474561753417224264885128710604089412672852950929900921658833506220956104517181
[/code][/QUOTE]

Line 2426 is 2^5 * 3 * 17 * 41839284258511 * 661383570738065494595663 * c119

Batalov 2009-07-28 08:02

[CODE]Input number is 44718209464695191784714835489610894962389758953872980040035617713029840660482876437561955501698845018995557237936968183 (119 digits)
Using B1=3000000, B2=5706890290, polynomial Dickson(6), sigma=3086004836
Step 1 took 7681ms
Step 2 took 4440ms
********** Factor found in step 2: 60743456023791205060853193364305989111
Found probable prime factor of 38 digits: 60743456023791205060853193364305989111
Probable prime cofactor 736181514716260899052084878748455443864989251384540365977037311428554914878692353 has 81 digits
[/CODE]

Batalov 2009-07-28 08:05

and then a c150...

Mini-Geek 2009-07-28 22:59

I just ran 100 curves at B1=3e6 on the c150, no luck. DB is showing 2156 curves at this B1 so far. GMP-ECM/aliqueit.exe suggest 2350 curves (so ~200 more) at this level before moving on, Syd's DB suggests 2500.
(the 2 curves reported at B1=3e7 were due to a typo in reporting :blush: they were not run)

axn 2009-07-28 23:44

[QUOTE=Mini-Geek;183200]I just ran 100 curves at B1=3e6 on the c150, no luck. DB is showing 2156 curves at this B1 so far. GMP-ECM/aliqueit.exe suggest 2350 curves (so ~200 more) at this level before moving on, Syd's DB suggests 2500.
(the 2 curves reported at B1=3e7 were due to a typo in reporting :blush: they were not run)[/QUOTE]

Suggest you directly move to 43e6.

Mini-Geek 2009-07-29 00:08

I'm not running any more ECM for the moment. Somehow I botched entering my curves into the DB. (most likely related to that I tried to set up a bookmark to submit curves...I know I didn't press it, maybe Firefox retrieves the URL at some point during my bookmark editing, without expecting it to mess something up?) I really only ran 20 curves at B1=11e6, not 100. Sorry for any confusion from that, but if we're moving straight to B1=43e6 it shouldn't be [I]too[/I] troublesome.

jrk 2009-07-29 02:28

+100 curves @ B1=43000000, B2=388112953420

10metreh 2009-07-29 07:21

Aliqueit suggests we should run just under t45. But I think 2*t45 might pay off.

Mini-Geek 2009-07-29 12:03

[quote=10metreh;183243]Aliqueit suggests we should run just under t45. But I think 2*t45 might pay off.[/quote]
What exactly does 2*t45 mean, in ECM?
Aliqueit suggests t44.65 (=.235*150+9.4), which is int(4480*(4.65/5))=4166 (does anyone know if this, ignoring the int() if necessary, is an exact probabilistic figure or just an approximation that aliqueit.exe uses?) curves at B1=11e6, so 2*t45 is...the 43e6 equivalent of 2*4166 curves at 11e6, maybe? What's that?

jrk 2009-07-30 07:51

+150 curves @ B1=43000000, B2=388112953420

jrk 2009-08-01 16:45

+145 curves @ B1=43000000, B2=388112953420

10metreh 2009-08-01 17:40

Come on guys! At this rate the ECMing will take longer than the sieving. :smile:

A t45 run is 1277 curves at 43e6. If we aim for t45, we have about 900 more curves to do (unless Serge's been running some in private).

Greebley 2009-08-03 13:10

I started a 1000 run for 43e6. Not sure how long it will take though as it is using only 1 processor.

Greebley 2009-08-04 04:26

Well 90 completed so 810 left. That will take me 5-7 days to complete without help.

Greebley 2009-08-04 18:25

I noticed we recently passed another milestone. This sequence is now at the highest number for seqs under 1 million(unless unknown progress has been made on 276). It passed 276 on this last index.

10metreh 2009-08-04 18:29

[QUOTE=Greebley;184086]I noticed we recently passed another milestone. This sequence is now at the highest number for seqs under 1 million(unless unknown progress has been made on 276). It passed 276 on this last index.[/QUOTE]

The highest we have the file for. 564 and 1134 are at 165 and 164 digits respectively, but the available files don't go that far.

Mini-Geek 2009-08-05 13:59

I just began 50 curves at 43e6 split across my two cores. Not sure how long it'll take. Edit: Step 1 takes about 10 minutes Edit 2: and step 2 takes about 225 seconds, so the 50 curves should be done in ~5.7 hours.

Mini-Geek 2009-08-05 19:54

[quote=Mini-Geek;184222]I just began 50 curves at 43e6 split across my two cores.[/quote]
Finished and in DB. (sadly, no factors found)

Greebley 2009-08-06 02:16

With Mini-Geek's work, we are now 500 runs away from finishing the full 900. It will take a bit less than 4 days for me to complete without help.

bsquared 2009-08-06 03:32

[quote=10metreh;183666]Come on guys! At this rate the ECMing will take longer than the sieving. :smile:

A t45 run is 1277 curves at 43e6. If we aim for t45, we have about 900 more curves to do (unless Serge's been running some in private).[/quote]

This is a little more than t45, but 2000 curves at 43e6 fits well in an overnight run, so those should be done in the morning. Sorry for not jumping in sooner... it's been more hectic than usual.

- b.

10metreh 2009-08-06 06:45

[QUOTE=bsquared;184303]This is a little more than t45[/QUOTE]

The 1277 figure is what I got from GMP-ECM 6.2.3.

bsquared 2009-08-06 12:28

[quote=bsquared;184303]This is a little more than t45, but 2000 curves at 43e6 fits well in an overnight run, so those should be done in the morning. Sorry for not jumping in sooner... it's been more hectic than usual.

- b.[/quote]

Done, no factor.

10metreh 2009-08-06 12:32

GNFS time?

bsquared 2009-08-06 13:17

Best so far in pol51, 0 to 100

[CODE]
BEGIN POLY #skewness 2948934.99 norm 2.49e+21 alpha -5.63 Murphy_E 3.65e-12[/CODE]

I'll take 100 to 1000 next for pol51

bsquared 2009-08-06 14:52

A couple more candidates from 100-1000

[CODE]seq4788_2428_550-775.cand:BEGIN POLY #skewness 1138330.08 norm 6.29e+20 alpha -5.84 Murphy_E 4.39e-12
seq4788_2428_775-1000.cand:BEGIN POLY #skewness 442881.85 norm 6.93e+20 alpha -5.68 Murphy_E 4.07e-12
seq4788_2428_775-1000.cand:BEGIN POLY #skewness 829132.56 norm 6.56e+20 alpha -5.33 Murphy_E 3.97e-12
[/CODE]

This is going fairly fast, so I'll take 1000-10000 next.

10metreh 2009-08-06 16:53

Ben, do you want to do the whole poly search?

bsquared 2009-08-06 17:43

[quote=10metreh;184352]Ben, do you want to do the whole poly search?[/quote]

Sure. It should be done by tomorrow.

[edit]
Although, I'm just planning on running pol51 tools... if anyone wants to chip in with msieve go ahead.

Greebley 2009-08-07 02:39

Checked my run and it found nothing as well and then I stopped it. Glad we don't have to sit around for 4 days waiting for the silly run to finish. 1 processor is slow at this level

bsquared 2009-08-07 13:44

1 Attachment(s)
After searching most of the way to 20000 and doing some test sieving, this poly was the clear winner:

[CODE]BEGIN POLY #skewness 621662.95 norm 5.35e+20 alpha -6.62 Murphy_E 5.31e-12[/CODE]

After some light optimization, these seem to be good parameters:

[CODE]
n: 601570829733431775831009462944836307372504373793253939901061852870883979386533253820339653192192369661940139258976230381609972509552130495736902586517
#BEGIN POLY #skewness 621662.95 norm 5.35e+20 alpha -6.62 Murphy_E 5.31e-12
type: gnfs
skew: 621662.95
c5: 4258980
c4: 9488946150708
c3: -5789674939355051089
c2: -3300143733276422915863786
c1: 830190772195714134449151313016
c0: 124539122235380375270696286356597376
Y1: 46807267200234637
Y0: -42657561083941041551403394433
lpbr: 28
lpba: 28
mfbr: 56
mfba: 56
rlim: 20000000
alim: 20000000
rlambda: 2.6
alambda: 2.6
[/CODE]

With rough estimates showing 3.3 million CPU seconds (38 CPU days) and 25 million Q needed (assuming 15% duplication rate). The 29 bit case was nearly a wash, so I picked the lower FTP overhead 28 bit case.

jrk 2009-08-07 14:30

Using siever 14e I assume?

Looks good. It is ready for sieving.

bsquared 2009-08-07 14:34

[quote=jrk;184453]Using siever 14e I assume?

Looks good. It is ready for sieving.[/quote]

Yep, sorry forgot to mention that.

We'll probably need to do Q's from say 5M to 30M or more. I'll start with 5M - 10M.

Oh, and I can do the post-processing for this one.

mdettweiler 2009-08-07 14:55

FTP server is open for business again, directory c150-relations. :smile:

10metreh 2009-08-13 09:46

We have a C123. Serge, is this yours?

Batalov 2009-08-13 09:55

No. Not taking.
And no. Not mine - someone, probably Ben, found a p36 factor, so it is ECM'd quite enough. I did add some curves but no luck (I must have spent it elsewhere).

bsquared 2009-08-13 12:49

I can't claim the p36... it was all I could do to post the factors before heading off to bed.

But I'll take the C123 unless someone else really wants it.

bsquared 2009-08-14 13:06

Seems I can't put these in the database right now... could someone do that when it comes back up? I'm leaving for a few days and won't be able to.

[CODE]prp54 factor: 162987771789684258473910202934761456347012754686995779
prp70 factor: 3476459109608271666227966958815165533599668181825613725957567035782921
[/CODE]

thanks,
- ben.

Mini-Geek 2009-08-14 13:25

[quote=bsquared;185521]Seems I can't put these in the database right now... could someone do that when it comes back up? I'm leaving for a few days and won't be able to.

[code]prp54 factor: 162987771789684258473910202934761456347012754686995779
prp70 factor: 3476459109608271666227966958815165533599668181825613725957567035782921
[/code]thanks,
- ben.[/quote]
Sure thing. :smile:
Anyone know what comes next for 4788? I don't have 4788's elf or the other cofactors in this line.
Edit: I noticed that the data from Greebley ended at 2427. I put in the factors for the C150 from 2428, but can't break the c133 on the next line very easily (it ran a few runs at 250k with no success, and I didn't want to waste too much work when someone probably had the data somewhere).

axn 2009-08-14 13:31

[QUOTE=Mini-Geek;185523]Sure thing. :smile:
Anyone know what comes next for 4788? I don't have 4788's elf or the other cofactors in this line.[/QUOTE]

The last c150 was part of a 2^5*3*c150. You can find the details from the relevant thread. I don't know how many other lines where there before the c123

10metreh 2009-08-14 18:25

[QUOTE=Mini-Geek;185523]Sure thing. :smile:
Anyone know what comes next for 4788? I don't have 4788's elf or the other cofactors in this line.
Edit: I noticed that the data from Greebley ended at 2427. I put in the factors for the C150 from 2428, but can't break the c133 on the next line very easily (it ran a few runs at 250k with no success, and I didn't want to waste too much work when someone probably had the data somewhere).[/QUOTE]

The c133 was p41*p92. The next line was small factors*p36*the c123. Ben, wanna have a crack at these to find the values of p41 and p36? (I remember the digit lengths, but not the numbers)

jrk 2009-08-14 19:08

[QUOTE=Mini-Geek;185523]Sure thing. :smile:
Anyone know what comes next for 4788? I don't have 4788's elf or the other cofactors in this line.
Edit: I noticed that the data from Greebley ended at 2427. I put in the factors for the C150 from 2428, but can't break the c133 on the next line very easily (it ran a few runs at 250k with no success, and I didn't want to waste too much work when someone probably had the data somewhere).[/QUOTE]
[code]
2427. 3593508421020970286512808613384151465284352922213963925811855953320436985791694024310820649020712727332399106329804838426440857002299108939582481253192826195296 = 2^5 * 3 * 17 * 19 * 263 * 4325813 * 250993631093 * 7981122652039 * 50850442337831500057375729431561590531623096768464082571136988112906722414532436212391901836529756733228800523429558275099
2428. 6959984711637874705440686112758310535782935268128068798556294131514289112057301048098148516117159418602327505055700449622256748568511056253130123882078474924704 = 2^5 * 3 * 37 * 3257231 * 211583091426921250197446583560454596406151386691 * 8232643115090532013514350738188694142495501455897 * 345355615167159325280013895125653060899714762489093471
2429. 11803763616190282764272497616546184352195160157977610660329621812935514416776522498920562554378255920514805511673913681041902473186552146049461184316268412177760 = 2^5 * 3 * 5 * 797 * 5413805017 * 3949781434631 * 23261849643148144875279862980194062520021 * 62029892180281651740726743184684307034982549212851927620182287210172536822010300883693892363
2430. 25424744046928495988198657869907720036018860040340586782247307354150718596022685350714189138925344762651129397398914553897144091085107776128687819954561164978848 = 2^5 * 3 * 467404843433471284216546751533412357 * 162987771789684258473910202934761456347012754686995779 * 3476459109608271666227966958815165533599668181825613725957567035782921[/code]

10metreh 2009-08-14 19:13

Next line is:
[code]41315209076258805980822819038600045201318969211789936794591428435375396599518680455218805759776349083356322831874886489751735221707037749030973936818492662199712 = 2^5 * 3 * 17 * 29 * 12641 * 33107 * 291521 * c142[/code]
with [code]c142 = 7155182203298369448937942346165428347449795593016786201865194819149797582908139440202912832303431350932284942688546565074936912797117840985577[/code]

The c142 has had hardly any ECM, so may have factors below 25 digits.

Mini-Geek 2009-08-14 19:15

[quote=jrk;185573][code]...[/code][/quote]
Thank you. :smile:
[code]4788:2431 = 41315209076258805980822819038600045201318969211789936794591428435375396599518680455218805759776349083356322831874886489751735221707037749030973936818492662199712
= 2^5*3*17*29*33107*12641*291521*c142
c142=7155182203298369448937942346165428347449795593016786201865194819149797582908139440202912832303431350932284942688546565074936912797117840985577[/code]The c142 has had 74 curves@11e3 and 53 curves@5e4 with no success. I'm not working on it more right now.
Edit: 10metreh beat me to it, and with about as much ECM. Oh well.

jrk 2009-08-14 19:19

From line 2431:

[code]Using B1=250000, B2=128992510, polynomial Dickson(3), sigma=3423068688
Step 1 took 805ms
Step 2 took 482ms
********** Factor found in step 2: 70421951056571174298407161
Found probable prime factor of 26 digits: 70421951056571174298407161
Composite cofactor 101604430095248107007562678598950397409062296407427396953273974341582170484206305218451473255594781352991727576633457 has 117 digits
[/code]

jrk 2009-08-14 19:24

Who wants to do the c117?

henryzz 2009-08-14 20:29

Should i be surprised that you got to 2*t25 and missed a 26 digit factor?

10metreh 2009-08-15 07:05

[QUOTE=henryzz;185580]Should i be surprised that you got to 2*t25 and missed a 26 digit factor?[/QUOTE]

Not too surprised - I once did t25 and t30 and missed a p25.

jrk, have you fully ECM'd the c117? (Just in case you haven't, I'll run a few curves myself)

Edit: seen Serge's post, aborted.

Batalov 2009-08-15 07:07

I've run more than enough 3e6 curves. Do go ahead. Anyone?
(I am bound with 552150.)

Mini-Geek 2009-08-15 12:44

About how long would that c117 take on two cores of my Athlon 2.5 GHz? (using factMsieve.pl, since I don't know enough to do it manually) If it's not too long, (either for my or this forum's patience) I'll take it. :smile: For comparison, a c109 with the same setup took about 7.5 hours for the GNFS.

Mini-Geek 2009-08-15 14:56

Ah, why not. I'm starting the GNFS now.

Mini-Geek 2009-08-15 19:14

status update: poly selection took 2:46:46 (h:mm:ss), first q=100k chunk (started at 1.8M) took about 66 minutes and found 772526 relations, and it wants at least 5737340 before filtering. (this is factMsieve.pl with the MINRELS factor changed from 0.2 to 0.4) If I find this many relations each time, it'll take about 8 hours before the first factoring run. I doubt it'll work with that few relations, though. A c109 I ran the other day took 7168224 before being solvable. I'll guestimate that it'll need 10M relations (anyone got a better approximation?). This puts it at 13 runs, or about 13 hours. Plus an hour or two for finishing. So maybe 15 hours total.
So I should have the factors when I wake up tomorrow morning, (about 6 AM CDT / 11 AM GMT) unless I'm grossly wrong about anything. (I'm safe by about 3 hours, by my estimates)

Mini-Geek 2009-08-16 12:16

[code]prp48 factor: 101787819775835948918733316910809569438934603559
prp69 factor: 998198314091099356256123369104680179895991846742675659120420684192423
[/code]It took 29 hours according to the computed timescale, just under 19 hours realtime (is the discrepancy due to the partial dual-threaded work or something?). I had 9602817 rels, which filtered down to 2292292 rels (if I understand the log correctly). The sieving was finished at about 3 AM my time, and the factors appeared at 4:47 AM.
Line 2432 is 2^5*3*p32*p128 and was already known as of when I put in the factors for the c117. Did someone beat me to the c117 and the next line without reporting the c117?
Anyway, now we're on Line 2433 and we lost a power of 2! :w00t: And gained a digit. :sad:
Cofactor is [URL="http://factordb.com/search.php?id=68093978"]c145[/URL]. Not much ECM has been done yet, (AFAIK, but the one who found that p32 may have worked on this c145 as well) but I'm keeping my status updated in the DB. :smile: Currently running 430@25e4. Just in case the DB has any more problems, here is the status:[code]2431. 41315209076258805980822819038600045201318969211789936794591428435375396599518680455218805759776349083356322831874886489751735221707037749030973936818492662199712 = 2^5 * 3 * 17 * 29 * 12641 * 33107 * 291521 * 70421951056571174298407161 * 101787819775835948918733316910809569438934603559 * 998198314091099356256123369104680179895991846742675659120420684192423
2432. 77489885742951318531011281537865398594576052855381059201245310474765450507848421490566102008207088378549525045932279226875927249701083945747558616485194709781088 = 2^5 * 3 * 28280105441052726736468878788089 * 28542549514354358893458915993662819655625753028398499312865912186369663866315908702072428978933749287812332066919759250319256377
2433. 125921064332295892612893332499038465438663703188435372848854032552047074765016841343996757973207483770437289808632875995776073092711376690665933668709679310599952 = 2^4 * 3 * 263 * 348407 * 6799769 * c145
c145 = 4210371029436746497837158825490947314990186258843761539738301861157299913528382593744449951063933548780440298195483979139600003150433674869074931
[/code]Edit: I finished 430@25e4 (t30) and some P-1 and P+1 (each with the highest B1 that Syd's DB would've assigned) unsuccessfully.

10metreh 2009-08-16 13:45

That must be the highest ever escape from 2^5 * 3! (161 digits)

frmky 2009-08-16 18:03

[QUOTE=Mini-Geek;185765]
Line 2432 is 2^5*3*p32*p128 and was already known as of when I put in the factors for the c117. Did someone beat me to the c117 and the next line without reporting the c117?
[/QUOTE]
Ok, I confess. I put it on the cluster to see where it would get overnight. The C117 GNFS took less than an hour, but since you were well into completing it yourself I couldn't bear to submit it. :ermm: Of course it got stuck on the C145. Overnight, the C145 survived 4500 curves at B1=3M. It automatically started the GNFS. If you want to do it as a team effort, I'll stop. Otherwise, I'll finish it by tomorrow.

henryzz 2009-08-16 18:27

[quote=frmky;185808]Ok, I confess. I put it on the cluster to see where it would get overnight. The C117 GNFS took less than an hour, but since you were well into completing it yourself I couldn't bear to submit it. :ermm: Of course it got stuck on the C145. Overnight, the C145 survived 4500 curves at B1=3M. It automatically started the GNFS. If you want to do it as a team effort, I'll stop. Otherwise, I'll finish it by tomorrow.[/quote]
As you have started it feel free to finish. Do you have a way of stopping it before it starts a large gnfs in the future(and putting on replacement work) when you are not around?

frmky 2009-08-16 18:39

[QUOTE=henryzz;185812]Do you have a way of stopping it before it starts a large gnfs in the future(and putting on replacement work) when you are not around?[/QUOTE]

Sure, just not start it. :grin:

unconnected 2009-08-16 20:04

Running 2000 curves @ 11e6 on C145 - no factors found.


All times are UTC. The time now is 14:47.

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