Aliqueit.exe discussion

[mdettweiler]

Originally Posted by mklasson

I think it's already taken care of. You should really get the fast 64-bit gnfs-lasieve if you're using a 64-bit machine though. Twice the speed, twice the fun.

Actually, even though I'm on a Core 2 Duo which is capable of 64-bit, I'm only running a 32-bit OS. It's primarily because this same machine used to have a Pentium 4 which I upgraded to the Core 2, and I never got around to doing a full reinstall of the OS. You're right, though, I really should get around to it sometime or other.

Anyway, thanks--glad to know I'm at least operating at full speed (at least by 32-bit standards). I should probably do some sec./rel comparisons between the Windows and Linux binaries, too, to see if the Windows one is operating at full speed as well.

[mklasson]

Well I'm off for a week. May all your sequences terminate in the meantime!

Happy factoring,
Mikael

[hhh]

I got ggnfs, the unxutils, adjusted the paths. aliqueit started selecting a polynomial, started sieving, and called msieve. So far so good; or so bad, because msieve refuses, Nicholas Nickleby-style, to do his duty:

Found 410524 relations, need at least 403406 to proceed.
=> "../msieve.exe" -s test.dat -l ggnfs.log -i test.ini -v -nf test.fb -t 2 -nc1


Msieve v. 1.38 for use in GGNFS 0.77.1-VC8(Sat 01/17/2009)

usage: ../msieve.exe [options]

options:
   -s <name> save intermediate results to <name>
             instead of the default msieve.dat
<snip>

Then: GGNFS failed to find a factor. Thisreally shouldn't happen, etc.

What is going on?

H.

[Andi47]

Originally Posted by hhh

I got ggnfs, the unxutils, adjusted the paths. aliqueit started selecting a polynomial, started sieving, and called msieve. So far so good; or so bad, because msieve refuses, Nicholas Nickleby-style, to do his duty:

Found 410524 relations, need at least 403406 to proceed.
=> "../msieve.exe" -s test.dat -l ggnfs.log -i test.ini -v -nf test.fb -t 2 -nc1


Msieve v. 1.38 for use in GGNFS 0.77.1-VC8(Sat 01/17/2009)

usage: ../msieve.exe [options]

options:
   -s <name> save intermediate results to <name>
             instead of the default msieve.dat
<snip>

Then: GGNFS failed to find a factor. Thisreally shouldn't happen, etc.

What is going on?

H.

I don't know what's going on, but you should upgrade to msieve 1.40. Msieve is used for poly selection (at least if you are using aliqueit 1.06), and msieve 1.39 or higher finds very good polynomials (very often far better than pol51), but msieve 1.38 or lower finds very bad polynomials.

[10metreh]

I called aliqueit with "aliqueit 130396 -e", trying to get it to run GNFS on a C95 that has cropped up. (I've just finished a C96, now another GNFS comes along!) It then spewed out "'msieve' is not recognized as an internal or external command, operable program or batch file". It has previously called msieve fine for QS, so why is it failing for NFS?

Just seen Mikael has been inactive for a week.

[Joshua2]

because its looking for msieve from the gnfs folder it spawned, its not there its up one folder. So you need to add msieve so it will always be every (forgetting temparily the proper name for that) or what I did fully reference msieve.exe in the folder, instead of just its name. Like c:/aliq/msieve.exe

[schickel]

Mikael, when starting a run on an existing file, can you have it just start if the partial factorization has correct factors?

I started a run on a file after aliquot.ub crashed (what better time?), but it had already written a partial line to sequence.txt.:

 841 .	 217333463898073716968749123915310163720522193994529887617661042267858643477340578271748105204 = 2^2 * 7

Obviously aliqueit.exe complained that the product didn't match the factors, but I would prefer that if a partial line like that is encountered, that it go ahead and resume the run. What about everyone else?

[Andi47]

When running multiple instances of aliqueit on a PC (e.g. a dual core), I think I have to create two different aliqueit folders to avoid logfile mismatches. What about naming the logfiles and temporary save files something like aliqueit<sequence_number>.log, factors<sequence_number>.log, ecm_<sequence_number>.out, msieve<sequence_number>.log etc.? (does yafu allow to specify how the output file is named?)

[mklasson]

Originally Posted by hhh

What is going on?
H.

I don't know what's going on either, but presumably the old msieve v1.38 doesn't accept the cmdline args ggnfs (factMsieve.pl) fed it. Like Andi said, try upgrading to msieve v1.40.

Originally Posted by 10metreh

I called aliqueit with "aliqueit 130396 -e", trying to get it to run GNFS on a C95 that has cropped up. (I've just finished a C96, now another GNFS comes along!) It then spewed out "'msieve' is not recognized as an internal or external command, operable program or batch file". It has previously called msieve fine for QS, so why is it failing for NFS?

Ailqueit creates a new directory for each ggnfs run so you'll need to put msieve.exe in a directory you've got included in your PATH env. var., or do like Joshua suggested and write the full path and filename to msieve.exe in aliqueit.ini.

Originally Posted by schickel

Mikael, when starting a run on an existing file, can you have it just start if the partial factorization has correct factors? [...] I would prefer that if a partial line like that is encountered, that it go ahead and resume the run. What about everyone else?

Yeah, I could do that if there's still any interest in it and/or I get some time to kill.

Originally Posted by Andi47

When running multiple instances of aliqueit on a PC (e.g. a dual core), I think I have to create two different aliqueit folders to avoid logfile mismatches. What about naming the logfiles and temporary save files something like aliqueit<sequence_number>.log, factors<sequence_number>.log, ecm_<sequence_number>.out, msieve<sequence_number>.log etc.? (does yafu allow to specify how the output file is named?)

Yeah, that's not a bad idea. I'll see what I can do.

[Joshua2]

Still you won't be able to call ecm or msieve again, right? I was getting that error when I tried two cores in one folder, it said ecm can't run and such; its not too bad to use 2 folders.

[Andi47]

Originally Posted by Joshua2

Still you won't be able to call ecm or msieve again, right? I was getting that error when I tried two cores in one folder, it said ecm can't run and such; its not too bad to use 2 folders.

It is at least possible to manually (i.e. without aliqueit) run two instances of ECM in the same folder - you just need to use different names for the input- and outputfiles. (I successfully did this quite often.)

It is also possible to run two instances of GGNFS in the same folder (with differently named output files) - I frequently do this manually, and also FactMsieve.pl is doing this if the number of threads is specified to be >1.

For msieve and yafu - I don't know.