Running GGNFS

[ET_]

Originally Posted by 10metreh

I have attached def-par.txt, which has estimates for parameters for C70-C140 GNFS and S100-S175 SNFS. Bear in mind that it is not exact, and for a GNFS C140, you might want to do some parameter optimization.

Thank you for the info

Luigi

[jasonp]

Originally Posted by ET_

As for Msieve, can it work with SNFS relations? If so, I will use nfs2ms.pl script after the sieving phase.

The only difference between SNFS and GNFS is that the polynomial in GNFS must be searched for (and the sieving parameters are different). Once you have relations, everything works the same between GNFS and SNFS. Msieve can handle either type.

[axn]

Originally Posted by ET_

I will reach 22 million relations, testing with Msieve after 15M, 18M and 21M, no problems.

Thanks for the hint, I just took the .poly from Kamada studio.
It seems that I'll have to improve my (low) knowledge on setting up better values. Links (apart from MersenneForum)?

Luigi

I've done some SNFS 176 digits myself. If you're lucky, you should be able to get away with as little as 14 million relations. I would suggest testing at 14m, and then every 1m relations. Oh, and factMsieve FTW

[ET_]

Originally Posted by axn

I've done some SNFS 176 digits myself. If you're lucky, you should be able to get away with as little as 14 million relations. I would suggest testing at 14m, and then every 1m relations. Oh, and factMsieve FTW

I will, thanks!

BTW, is there a version of factMsieve for linux environments?

Luigi

[schickel]

Originally Posted by ET_

Well, I had that same message every 100000 q, so I thought there was something missing in my environment.

If you had the message every 100k q and your spairs.out file has >5 million relations, something has been going wrong from the beginning.

As there don't appear to be any "rels.bin.*" files in your work directory, procrels has not been processing the relations after every block. Luckily the siever you're using does not delete the spairs.out file at the start of each block. (The early versions did.)

As for Msieve, can it work with SNFS relations? If so, I will use nfs2ms.pl script after the sieving phase.

Luigi

nfs2ms.pl dumps the relations from the "rels.bin.*" files prior to running msieve. No rels.bin.* files means no automatic post processing using nfs2ms.pl. You could probably fake it by making a copy of the spairs.out file named "spairs.dump.0" before running nfs2ms.pl. Procrels will complain about no rels.bin.0 file and die, but the formatting for msieve should take place. If you have the disk space, you could copy the directory to make sure you don't lose the relations you have while you experiment....

BTW, is there a version of factMsieve for linux environments?

As it's a perl script, it's platform independent.....

[ET_]

Originally Posted by schickel

If you had the message every 100k q and your spairs.out file has >5 million relations, something has been going wrong from the beginning.

As there don't appear to be any "rels.bin.*" files in your work directory, procrels has not been processing the relations after every block. Luckily the siever you're using does not delete the spairs.out file at the start of each block. (The early versions did.)nfs2ms.pl dumps the relations from the "rels.bin.*" files prior to running msieve. No rels.bin.* files means no automatic post processing using nfs2ms.pl. You could probably fake it by making a copy of the spairs.out file named "spairs.dump.0" before running nfs2ms.pl. Procrels will complain about no rels.bin.0 file and die, but the formatting for msieve should take place. If you have the disk space, you could copy the directory to make sure you don't lose the relations you have while you experiment....

As it's a perl script, it's platform independent.....

Yes, but AFAIK, its directory configuration uses Windows syntax and environment I didn't want to mess up with the script, as long as it works for Windows...

Luigi

[Jeff Gilchrist]

Originally Posted by ET_

Yes, but AFAIK, its directory configuration uses Windows syntax and environment I didn't want to mess up with the script, as long as it works for Windows...

If you grab the latest version of factMsieve.pl from:
http://ggnfs.svn.sourceforge.net/vie.../factMsieve.pl

The default directory settings are Linux syntax and environment. You have to go out of your way to make it work in Windows. Look at Step 2 of my guide to see the important sections to change in the script: http://gilchrist.ca/jeff/factoring/n...ers_guide.html

[ET_]

Originally Posted by Jeff Gilchrist

If you grab the latest version of factMsieve.pl from:
http://ggnfs.svn.sourceforge.net/vie.../factMsieve.pl

The default directory settings are Linux syntax and environment. You have to go out of your way to make it work in Windows. Look at Step 2 of my guide to see the important sections to change in the script: http://gilchrist.ca/jeff/factoring/n...ers_guide.html

I'll be running it from Linux, so the default script you pointed me on should work fine, thanks ;-)

Luigi

[wsgtrsys]

i Factoring rsa512.
and log is:

[02/20 21:21:32] GGNFS-0.77.1-20090220-pentium4 : makefb
[02/20 21:21:41] name: template
[02/20 21:21:41] n=1219901374107509269727705827206869949617579394785535062013138456055113298999855436482715704794764249901926
4099857141857117583668732368414756143066697037831 (155 digits)
[02/20 21:21:41] c0: -1654136390110887053567413679248552485
[02/20 21:21:41] c1: 6896128775024703671131698569170
[02/20 21:21:41] c2: 7591355774065335794017922
[02/20 21:21:41] c3: -23229658654233831899
[02/20 21:21:41] c4: -8643354678366
[02/20 21:21:41] c5: 4603320
[02/20 21:21:41] RFBsize: 374362 (upto 5399993)
[02/20 21:21:41] AFBsize: 374321 (upto 5399993)
[02/20 21:21:41] maxNumLargeRatPrimes: 3
[02/20 21:21:41] maxLargeRatPrime: 134217728
[02/20 21:21:41] maxNumLargeAlgPrimes: 3
[02/20 21:21:41] maxLargeAlgPrime: 134217728
-> minimum number of FF's: 838596
-> makeJobFile(): Adjusted to q0=2700000, q1=2760000.
-> client 1 q0: 2700000
-> minimum number of FF's: 838596
-> makeJobFile(): Adjusted to q0=2760000, q1=2820000.
-> client 2 q0: 2760000
-> minimum number of FF's: 838596
-> makeJobFile(): Adjusted to q0=2820000, q1=2880000.
-> client 3 q0: 2820000
-> minimum number of FF's: 838596
-> makeJobFile(): Adjusted to q0=2880000, q1=2940000.
-> client 4 q0: 2880000
-> minimum number of FF's: 838596
-> makeJobFile(): Adjusted to q0=2940000, q1=3000000.
-> client 5 q0: 2940000
-> minimum number of FF's: 838596
-> makeJobFile(): Adjusted to q0=3000000, q1=3060000.
-> client 6 q0: 3000000
-> minimum number of FF's: 838596
-> makeJobFile(): Adjusted to q0=3060000, q1=3120000.
-> client 7 q0: 3060000
-> minimum number of FF's: 838596
-> makeJobFile(): Adjusted to q0=3120000, q1=3180000.
-> client 8 q0: 3120000
-> minimum number of FF's: 838596
-> makeJobFile(): Adjusted to q0=3180000, q1=3240000.
-> client 9 q0: 3180000

There are 0 relations with 4 large primes.
There are 0 relations with 5 large primes.
There are 0 relations with 6 large primes.
Doing merge on chunk 1/1 (P0=0, P1=8449333)...
Doing 13 additions...
* There are now 5901 full relations.
pass 4...
Before sortByNumLP()... Doing ll_verify(P)...
ll_verify() reports that 'P' appears to be intact.
makePass:
There are 5901 relations with 0 large primes.
There are 13 relations with 1 large primes.
There are 0 relations with 2 large primes.
There are 0 relations with 3 large primes.
There are 0 relations with 4 large primes.
There are 0 relations with 5 large primes.
There are 0 relations with 6 large primes.
After sortByNumLP()... Doing ll_verify(P)...
ll_verify() reports that 'P' appears to be intact.
Deleting 13 singleton large primes.
Deleting 0 singleton large primes.
Total: 13 singletons deleted.
makePass:
There are 5901 relations with 0 large primes.
There are 0 relations with 1 large primes.
There are 0 relations with 2 large primes.
There are 0 relations with 3 large primes.
There are 0 relations with 4 large primes.
There are 0 relations with 5 large primes.
There are 0 relations with 6 large primes.
Doing merge on chunk 1/1 (P0=0, P1=0)...
mkLT: There appear to be no large primes in the specified range.
* There are now 5901 full relations.
After keepFulls(), R->numFields = 5901
More columns needed (current = 5901, min = 838596)
-> Found 5901 relation-sets versus minFF=838596.
-> More sieving needed.
-> Q0=118680001, QSTEP=60000.
-> makeJobFile(): q0=119460000, q1=119520000.
-> makeJobFile(): Adjusted to q0=119460000, q1=119520000.
-> Lattice sieving algebraic q-values from q=119460000 to 119520000.
=> "../../bin/gnfs-lasieve4I14e" -k -o spairs.out -v -n1 -a ros.job
FBsize 374316+0 (deg 5), 374361+0 (deg 1)
total yield: 2402, q=119496059 (3.36285 sec/rel)

i wan't know how much time can Factoring it?

[Batalov]

With a good choice of parameters, ~100 pentium4 cpu-days.
With a bad choice (or leaving the choice to the script without even knowing what it's doing), "it may take the term of your natural life".
(See 'A Beautiful Mind' for the origin of this quote).

[ATH]

Is there any list of all the switches/parameters for the GGNFS laticesievers?

I can't find it in ggnfs-doc.pdf or any of the other documents.