Discussion related to cado-nfs

[Paul Leyland <paul.leyland@gmail.com>]

OK, I think I see what may have gone wrong.  As I said, I'm a complete
newby.

To start with I used the 1.1 release and the factor.sh script but moved
to the git version after mailing the other Paul.  In the interim a power
failure killed the factorization and the reboot wiped /tmp which is
where all the siever output had been stored.  Much of /tmp was safely on
a back-up so I restored it and tried to restart with the git version.
Everything worked well until the problem just reported.

In the restored files were mach_desc and c147.param files which still
contained the path to the -1.1 binaries.  Those two files are now edited
to point to the latest path and the script restarted.  dup2 is running
now; with luck purge will take over next.

Your question was enough to draw my attention to the issue.

Thanks,
        Paul (this one)

On Thu, 2013-04-04 at 20:25 +0200, Cyril Bouvier wrote:
> Paul,
> 
> The output of purge in c147.purge.log suggest that this is not the
> latest version of purge that is executed.
> 
> Are you sure that the version of purge and the version of cadofactor.pl
> that are running are identical ?
> 
> > Should I let the run continue or should I interrupt it and try to
> > recover somehow?
> You can stop the factorization as the purge binary will always fails in
> the same way (unused parameter) and the sieving step will never stop.
> You can later recover the work that are already done by using
> cadofactor.pl with exactly the same arguments on the command line.
> 
> Cyril
> 
> 
> 
> [2013-04-04 17:51:39 +0100] Paul Leyland:
> > I'm running the latest git repository version of CADO-NFS on a c147 and
> > was wondering why I never had enough relations.  As I'm a complete newby
> > I may genuinely not have enough, but I also have lines like this in
> > purge.log
> > 
> > pcl@maat ~/cado-nfs/test $ cat tmp/c147.purge.log
> > /home/pcl/cado-nfs-1.1/build/maat/filter/purge.ra55366a
> > -poly ./tmp//c147.poly -keep 160 -nrels 38129246
> > -out ./tmp//c147.purged.gz -npthr 2x2 -basepath ./tmp/
> > -subdirlist ./tmp//c147.subdirlist -filelist ./tmp//c147.filelist
> > Warning: unused command-line parameter npthr
> > Usage: purge [options] -poly polyfile -out purgedfile -nrels nnn
> > [-basepath <path>] [-subdirlist <sl>] [-filelist <fl>] file1 ... filen
> > Options:
> >        -excess  nnn - initial excess (default 1.01)
> >        -keep    nnn - prune if excess > nnn (default 160)
> >        -minpa   nnn - purge alg. primes >= nnn (default alim)
> >        -minpr   nnn - purge rat. primes >= nnn (default rlim)
> >        -nprimes nnn - expected number of prime ideals
> >        -sos sosfile - to keep track of the renumbering
> >        -raw         - output relations in CADO format
> > 
> > 
> > Should I be worried?
> > 
> > Should I let the run continue or should I interrupt it and try to
> > recover somehow?
> > 
> > Right now there are "Info:Running total: 47706267 relations." which is
> > plausibly large enough to complete the factorization.
> > 
> > The factorization was started with
> > 
> > ../cadofactor.pl params=../params/params.c147 wdir=./tmp/ name=c147 
> > n=`cat num.n`
> > 
> > and, of course, num.n contains a single line containing 147 decimal digits
> > 
> > Thanks,
> >     Paul
> > 
> > 
> > 
> > 
> > _______________________________________________
> > Cado-nfs-discuss mailing list
> > Cado-nfs-discuss@lists.gforge.inria.fr
> > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/cado-nfs-discuss