- I have compiled GGNFS using the default settings (Makefile ->
Makefile.athlon) and even successfully factored a SNFS difficulty 165
number on my first try. (It took about 82 hours on a 1.8GHz T2400
running on a single core.)
I am using the following source code:
I wanted to see if the x86 version was faster so I set Makefile ->
Makefile.x86 and compiled it, but GGNFS stops execution when it tries
to invoke matbuild.exe because matbuild.c is not compiled in the
Makefile.x86 make file.
I just copied over the matbuild.exe from my original build for now and
it works fine.
I guess I'm posting this as sort of a bug report, but more likely I am
just doing something wrong.
BTW, I have tested this on both a Debian system and a Cygwin system.
- During my current factorization of a c137 the siever stops quite
frequently with a return value 65280.
Restarting with the same specialq returns the same value. I have to skip
to a higher specialq before i'm able to continue.
I'm using about 10 cpu's and several times a day i have to check if they
are still running.
=> "../../bin/gnfs-lasieve4I14e.exe" -k -o spairs.out -v -n1 -a in.job
FBsize 584794+0 (deg 5), 583713+0 (deg 1)
total yield: 55546, q=10386029 (0.30845 sec/rel)-> Return value
65280. Updating job file and terminating...
=>"../../bin/cat.exe" spairs.out >> spairs.add
I've seen this on more factorizations, but this one is the worst.
What causes this problem and is there a way to prevent them (like
letting factlat.pl skip 10K specialq and continue?