Factoring with GGNFS

[VolMike]

I got GGNFS from http://sourceforge.net/project/showf...ease_id=416680
(source code and binaries for P4).Tex document doesn't help me to find the answer.
In what sequence and with which flags should I write executable files in *.bat file to factorize number,given in *.n format files(f.e. rsa110.n)?

VolMike

[smh]

Originally Posted by VolMike

I got GGNFS from http://sourceforge.net/project/showf...ease_id=416680
(source code and binaries for P4).Tex document doesn't help me to find the answer.
In what sequence and with which flags should I write executable files in *.bat file to factorize number,given in *.n format files(f.e. rsa110.n)?

VolMike

Take a look at the factlat.pl script. I can do the complete factorization for you.

You can use activeperl to run the script on windows.

[VolMike]

Originally Posted by smh

Take a look at the factlat.pl script. I can do the complete factorization for you.

You can use activeperl to run the script on windows.

I was trying to use script, but there were some errors (some method of some class wasn't found) while compiling with perl command, that's why I want to use binaries, which are not depends on classes,methodes,libraries and so on.

[smh]

Originally Posted by VolMike

I was trying to use script, but there were some errors (some method of some class wasn't found) while compiling with perl command, that's why I want to use binaries, which are not depends on classes,methodes,libraries and so on.

No idea what you're talking about with compiling the script. I just run the scrip, which works fine for me.

If you can't run perl scripts, you can use it to see what programs to run and what parameters to use.

Use POL51M0 and POL51OPT to find a polynomial (not needed for SNFS where you create the poly your self).

Then makefb for creating the factor base, gnfs-lasieve for sieving followed by procrels, matbuild, matsolve and finally sqrt

[antiroach]

I've got a question about GGNFS. Im using the factLat.pl script. The QSTEP value is chosen as 100,000. Every 100,000 q sieved the procrels and matbuild programs also execute. At the start this isnt a big deal because they dont take long to execute. However now that I have about 600MB worth of relations those steps take an increasing amount of time.

I noticed i can just increase the QSTEP value in the .poly file, but then I dont want to run into problems with having the spairs.out file getting too large and cause problems.

My question is, is it safe to increase the QSTEP size and have the spairs.out file grow (upwards of 2GB lets say) before the procrels and matbuild programs run?

[antiroach]

i've got another question. every once in a while i get this warning message:

warning: too many relations in mpqs

is this something that can be ignored or is it a problem ?

[jasonp]

Originally Posted by antiroach

warning: too many relations in mpqs

is this something that can be ignored or is it a problem ?

It can be ignored; a part of the lattice siever tries to factor small numbers, and there is a maximum effort it can expend to do so. When that maximum is exceeded the relation it was working on gets thrown away, but it's just one relation.

[antiroach]

I recently completed (?) my first factorization with GGNFS. It was of a c109 from the partition numbers list posted in another thread. However it doesnt seem that it actually found any factors. I've attached the two relevant (I hope) log files. I used the factLat.pl script. Is it possible that it didnt find enough relations ? If so how can i force it to compute more relations?

[antiroach]

Here's the other file since the upload limit is 1.

[fivemack]

The factorisation failed at the square-root stage; you can tell you had enough relations because the 'matsolve' program was used, which means the relations did get boiled down into a matrix.

My recipe at this point is

* Download msieve-1.25 (see the jasonp thread in this forum)
* Run 'procrels -prel rels.bin -fb {factor base filename} -dump'
* Use the make-msieve perl script that I posted in an earlier thread to convert the many spairs.dump.* files into a format that msieve can handle
* Run msieve to finish the factorisation

[antiroach]

i will try this. thanks.