[QUOTE=xilman;112988]You're correct.

No-one seems to know why this phenomenon occurs, but it undoubtedly does and it does so in several independently developed implementations.

Paul[/QUOTE]

Someday when I have the time, it would be interesting to try my own implementation on these examples to see if I get the same behviour.


Chris

I'm sieving a c137 with GGNFS at the moment. Since i'm planning to finish the factorization with msieve i took the relations so far to see how much more sieving was needed.

GGNFS reports:[code][09/10 12:48:47] largePrimes: 10291902 , relations: 11469979[/code]
while msieve reports [code]Mon Sep 10 20:35:35 2007 found 147506 hash collisions in 2815350 relations[/code] What happened to my other 8,6M relations?

Full msieve.log below.[code]Mon Sep 10 20:34:15 2007
Mon Sep 10 20:34:15 2007
Mon Sep 10 20:34:15 2007 Msieve v. 1.26
Mon Sep 10 20:34:15 2007 random seeds: 886d2c28 8dfb06ca
Mon Sep 10 20:34:15 2007 factoring 33553602168194917406374501182161257780336905415860676987778951034243560154498095405935359653092114506393135863732719626541228975712026099 (137 digits)
Mon Sep 10 20:34:16 2007 commencing number field sieve (137-digit input)
Mon Sep 10 20:34:16 2007 R0: -124431777510848056439439144
Mon Sep 10 20:34:16 2007 R1: 583622621142187
Mon Sep 10 20:34:16 2007 A0: -34268730864255576977750515205855
Mon Sep 10 20:34:16 2007 A1: 885939207896215753745532591
Mon Sep 10 20:34:16 2007 A2: 5497401588291985238782
Mon Sep 10 20:34:16 2007 A3: -84045614687373385
Mon Sep 10 20:34:16 2007 A4: -223188673593
Mon Sep 10 20:34:16 2007 A5: 1124820
Mon Sep 10 20:34:16 2007 size score = 4.963706e-014, Murphy alpha = -5.111970, combined = 2.727969e-013
Mon Sep 10 20:34:21 2007 restarting with 2815364 relations
Mon Sep 10 20:34:21 2007 generating factor base
Mon Sep 10 20:34:33 2007 factor base complete:
Mon Sep 10 20:34:33 2007 583714 rational roots (max prime = 8699993)
Mon Sep 10 20:34:33 2007 584796 algebraic roots (max prime = 8699993)
Mon Sep 10 20:34:35 2007 added 4938 free relations
Mon Sep 10 20:34:35 2007
Mon Sep 10 20:34:35 2007 commencing relation filtering
Mon Sep 10 20:34:35 2007 commencing duplicate removal, pass 1
Mon Sep 10 20:34:43 2007 error -14 reading relation 373948
Mon Sep 10 20:34:45 2007 error -14 reading relation 470959
Mon Sep 10 20:34:45 2007 error -10 reading relation 475156
Mon Sep 10 20:34:51 2007 error -8 reading relation 747056
Mon Sep 10 20:34:52 2007 error -14 reading relation 799021
Mon Sep 10 20:34:53 2007 error -14 reading relation 806668
Mon Sep 10 20:34:56 2007 error -14 reading relation 999537
Mon Sep 10 20:34:59 2007 error -14 reading relation 1103640
Mon Sep 10 20:35:00 2007 error -4 reading relation 1149143
Mon Sep 10 20:35:08 2007 error -1 reading relation 1557955
Mon Sep 10 20:35:13 2007 error -14 reading relation 1783111
Mon Sep 10 20:35:15 2007 error -14 reading relation 1881457
Mon Sep 10 20:35:21 2007 error -14 reading relation 2183618
Mon Sep 10 20:35:23 2007 error -8 reading relation 2273453
Mon Sep 10 20:35:35 2007 found 147506 hash collisions in 2815350 relations
Mon Sep 10 20:35:35 2007 commencing duplicate removal, pass 2
Mon Sep 10 20:35:48 2007 found 134269 duplicates and 2681081 unique relations
Mon Sep 10 20:35:48 2007 memory use: 37.8 MB
Mon Sep 10 20:35:55 2007 ignoring smallest 401129 rational and 401843 algebraic ideals
Mon Sep 10 20:35:55 2007 filtering rational ideals above 5817624
Mon Sep 10 20:35:55 2007 filtering algebraic ideals above 5817624
Mon Sep 10 20:35:55 2007 need 1365052 more relations than ideals
Mon Sep 10 20:35:55 2007 commencing singleton removal, pass 1
Mon Sep 10 20:36:54 2007 relations with 0 large ideals: 24480
Mon Sep 10 20:36:54 2007 relations with 1 large ideals: 194328
Mon Sep 10 20:36:54 2007 relations with 2 large ideals: 620174
Mon Sep 10 20:36:54 2007 relations with 3 large ideals: 941544
Mon Sep 10 20:36:54 2007 relations with 4 large ideals: 681452
Mon Sep 10 20:36:54 2007 relations with 5 large ideals: 197369
Mon Sep 10 20:36:54 2007 relations with 6 large ideals: 20752
Mon Sep 10 20:36:54 2007 relations with 7+ large ideals: 982
Mon Sep 10 20:36:54 2007 2681081 relations and about 4808135 large ideals
Mon Sep 10 20:36:54 2007 commencing singleton removal, pass 2
Mon Sep 10 20:37:54 2007 found 2296937 singletons
Mon Sep 10 20:37:54 2007 current dataset: 384144 relations and about 586342 large ideals
Mon Sep 10 20:37:54 2007 commencing singleton removal, pass 3
Mon Sep 10 20:38:07 2007 found 309525 singletons
Mon Sep 10 20:38:07 2007 current dataset: 74619 relations and about 54198 large ideals
Mon Sep 10 20:38:07 2007 commencing singleton removal, final pass
Mon Sep 10 20:38:13 2007 memory use: 10.9 MB
Mon Sep 10 20:38:13 2007 commencing in-memory singleton removal
Mon Sep 10 20:38:13 2007 begin with 74619 relations and 55006 unique ideals
Mon Sep 10 20:38:13 2007 reduce to 35061 relations and 5895 ideals in 6 passes
Mon Sep 10 20:38:14 2007 max relations containing the same ideal: 6
Mon Sep 10 20:38:14 2007 filtering wants 4007658 more relations
Mon Sep 10 20:38:14 2007 c137 factor: 33553602168194917406374501182161257780336905415860676987778951034243560154498095405935359653092114506393135863732719626541228975712026099
Mon Sep 10 20:38:14 2007 elapsed time 00:03:59
[/code]

[QUOTE=smh;113994]What happened to my other 8,6M relations?[/QUOTE]Never mind. I let factlat.pl save the relations (using gzip) and unzipped them with winzip and used the copy command to add the number to the first line (is there an easier way to do this?).

Something probably went allong the way.

After dumping all relations msieve finds them all, except for about 40.000 error -14 reading relation xxxx messages.

[QUOTE=smh;114007]Never mind. I let factlat.pl save the relations (using gzip) and unzipped them with winzip and used the copy command to add the number to the first line (is there an easier way to do this?).

Something probably went allong the way.

After dumping all relations msieve finds them all, except for about 40.000 error -14 reading relation xxxx messages.[/QUOTE]
40,000 relation errors is really a lot. I know that you'll get error -14 if the b value of the relation is larger than 2^31 (this should be fixed in GGNFS), but that usually is only an issue with SNFS and usually only kills ~10 relations.

I am running Msieve on a C114, but when I went to the task manager, msieve is only using 50% of the computer resources, while the total usage is ~51%.

Is there some way I can run a second instance of msieve to speed up the factorization? For example, could I type "msieve -s C114.dat -nf C114.dat" for both of them? Or would I have to make two .dat files and combine them afterwards?

Thanks,

Roger

[QUOTE=roger;114196]
Is there some way I can run a second instance of msieve to speed up the factorization? For example, could I type "msieve -s C114.dat -nf C114.dat" for both of them? Or would I have to make two .dat files and combine them afterwards?
[/QUOTE]
You have to generate two data files and manually combine them afterwards. I couldn't find a portable way to lock the datafile so that multiple writers could append to it safely.

Okay, thanks :smile:

is there a list of all the parameters for msieve
also when is the next version coming out

[QUOTE=henryzz;114331]is there a list of all the parameters for msieve
also when is the next version coming out[/QUOTE]
If you want a list of command line options try 'msieve -h'. If you want the parameters given to the actual factoring code, there's no comprehensive list.

The next version will be out when I can manage improved NFS polynomial selection, which will probably take months. If someone finds a nasty bug, the next version will come out sooner.

Hello

Somewhere in the mersenneforum, I heard about a program to combine large files. I am factoring a C114 with msieve, and so far have two files of 75Mb each. I would like to combine them so that I don't have to do the factorization twice, but Notepad and wordpad (Windows XP) can't take it.

Where can I find this program?

Thanks,

Roger

[QUOTE=roger;114448]Hello

Somewhere in the mersenneforum, I heard about a program to combine large files. I am factoring a C114 with msieve, and so far have two files of 75Mb each. I would like to combine them so that I don't have to do the factorization twice, but Notepad and wordpad (Windows XP) can't take it.

Where can I find this program?

Thanks,

Roger[/QUOTE]Open a command prompt in the directory with the sieve files and use the copy command:

copy /B msieve1.dat+msieve2.dat msieve.dat