Yafu

[mnh001]

I don't believe so. The zip file I got from sourceforge was

yafu-1.24.zip

then from Jeff Christie's? page I got the zips

ggnfs-svn413-win64-core2.zip ...Oh, I see. I got the wrong support file. I didn't even notice that. Thanks. I'll go find the correct files and try again. :)

[mnh001]

Well, apparently using the proper files didn't help much. I used
ggnfs-svn413-win32-p4.zip
but it still crashed. The log shows:

04/03/11 11:37:11 v1.24 @ HOME, nfs: commencing gnfs on c100: 1018979337456225035714934025075344288205054863323166920696334625238968374307490538496140762051312209
04/03/11 11:37:11 v1.24 @ HOME, nfs: commencing NFS restart - searching for last special-q
04/03/11 11:37:11 v1.24 @ HOME, nfs: continuing job - could not determine last special q; using default startq
04/03/11 11:37:11 v1.24 @ HOME, nfs: commencing lattice sieving with 1 threads
04/03/11 11:37:11 v1.24 @ HOME, could not open rels0.dat for reading

Also, in the main YAFU folder I do have a ggnfs.job file but the icon has a red X thru it and I can't open it. I tried to attach it here but got a message saying it was an invalid file. Any suggestions?

[Brian Gladman]

Did you clean out all the old data files before running the 32-bit binaries? Maybe some files left over from earlier attempts are getting in the way?

[mnh001]

The only files that look like they were created by runtime are

factor.log, ggnfs.job, msieve.log, profilingData.log, and session.log

so I moved them all elsewhere and will now try again.

Oh, also, all the old files from the 64-bit zip I deleted before starting with the 32-bit.

[mnh001]

Ok, tried again, same crash. I've got the win32 files. There are no old files to interfere. Here's the end of the factor log:

04/03/11 14:24:46 v1.24 @ HOME, Finished 1 curves using Lenstra ECM method on C100 input, B1 = 10M, B2 = gmp-ecm default
04/03/11 14:24:46 v1.24 @ HOME, nfs: commencing gnfs on c100: 1018979337456225035714934025075344288205054863323166920696334625238968374307490538496140762051312209
04/03/11 14:24:46 v1.24 @ HOME, nfs: commencing msieve poly select
04/03/11 14:44:52 v1.24 @ HOME, nfs: commencing lattice sieving with 1 threads
04/03/11 14:44:52 v1.24 @ HOME, could not open rels0.dat for reading

During the run it created: factor.log, msieve.log, session.log, ggnfs.job (with a red X thru the icon), msieve.dat and msieve.fb.

rels0.dat was not created.

As for independent test run of gnfs-lasieve4I12e.exe, when I double click it I very briefly (~0.1 sec) see the window flash on the screen then it's gone.

Now what?

[mnh001]

Oh, I don't know if you need this or not but here's the contents of msieve.log:

Sun Apr 03 14:24:46 2011 commencing number field sieve (100-digit input)
Sun Apr 03 14:24:46 2011 commencing number field sieve polynomial selection
Sun Apr 03 14:24:46 2011 time limit set to 0.33 hours
Sun Apr 03 14:24:46 2011 searching leading coefficients from 1 to 17810677
Sun Apr 03 14:44:52 2011 polynomial selection complete
Sun Apr 03 14:44:52 2011 R0: -1435442859724036989108407
Sun Apr 03 14:44:52 2011 R1: 26386304610883
Sun Apr 03 14:44:52 2011 A0: 36621874029161643309794237076
Sun Apr 03 14:44:52 2011 A1: -4389786114023453402424
Sun Apr 03 14:44:52 2011 A2: -26705947173531107
Sun Apr 03 14:44:52 2011 A3: 316114064
Sun Apr 03 14:44:52 2011 A4: 240
Sun Apr 03 14:44:52 2011 skew 5931426.95, size 1.241e-013, alpha -5.044, combined = 1.319e-008 rroots = 4

[bsquared]

Originally Posted by mnh001

Now what?

Two things.

1) open a command prompt, navigate to the directory where ggnfs.job is located, and enter the following command:

type ggnfs.job

Post whatever is returned by this.

2) If the ggnfs.job file is not empty, enter the following command:

<your-directory-here>\gnfs-lasieve4I12e.exe -a ggnfs.job -f 1000000 -c 100 -o test.dat

where you've substituted the absolute path to your binaries as indicated. If everything works, you should see something like the following:

 Warning:  lowering FB_bound to 999999.
total yield: 242, q=1000117 (0.00444 sec/rel)

If so, post it. If all of this works, then there might be something wrong with yafu... but right now I still suspect faulty gnfs-lasieve binaries or corrupted files.

[mnh001]

Ok, I did it. Couldn't remember my DOS very well so I copied the files to C:\ first and did it from there.

For the contents of ggnfs.job :

C:\>type ggnfs.job
n: 10189793374562250357149340250753442882050548633231669206963346252389683743074
90538496140762051312209
skew: 5931426.95
Y0: -1435442859724036989108407
Y1:  26386304610883
c0:  36621874029161643309794237076
c1: -4389786114023453402424
c2: -26705947173531107
c3:  316114064
c4:  240
rlim: 1800000
alim: 1800000
lpbr: 26
lpba: 26
mfbr: 52
mfba: 52
rlambda: 2.500000
alambda: 2.500000

For running gnfs-lasieve4I12e.exe I got :

C:\>c:\gnfs-lasieve4I12e.exe -a ggnfs.job -f 1000000 -c 100 -o test.dat
 Warning:  lowering FB_bound to 999999.
total yield: 877, q=1000117 (0.00210 sec/rel)

[bsquared]

Ok... try copying yafu-32k-Win32.exe yafu.ini and gnfs-lasieve4I12e into the same directory as the ggnfs.job file and change the line in yafu.ini to ggnfs_dir=.\

Make a copy of the ggnfs.job file, so that you don't have to redo the poly selection if something goes wrong. Then restart yafu nfs() with the same N.

[mnh001]

Ok I tried all that. Here's what I got:

screen output while running

Found expression: 1018979337456225035714934025075344288205054863323166920696334625238968374307490538496140762051312209
nfs: commencing gnfs on c100: 1018979337456225035714934025075344288205054863323166920696334625238968374307490538496140762051312209
nfs: continuing job - could not determine last special q; using default startq
nfs: commencing lattice sieving over range: 900000 - 940000
 Warning:  lowering FB_bound to 899999.
total yield: 12410, q=900743 (0.00215 sec/rel)

factor.log afterwards

04/03/11 20:39:26 v1.24 @ HOME, nfs: commencing gnfs on c100: 1018979337456225035714934025075344288205054863323166920696334625238968374307490538496140762051312209
04/03/11 20:39:26 v1.24 @ HOME, nfs: commencing NFS restart - searching for last special-q
04/03/11 20:39:26 v1.24 @ HOME, nfs: continuing job - could not determine last special q; using default startq
04/03/11 20:39:26 v1.24 @ HOME, nfs: commencing lattice sieving with 1 threads
04/03/11 21:00:52 v1.24 @ HOME, nfs: commencing msieve filtering

I'm not sure if it used the ggnfs.job but here it is after run

C:\>type ggnfs.job
n: 10189793374562250357149340250753442882050548633231669206963346252389683743074
90538496140762051312209
skew: 5931426.95
Y0: -1435442859724036989108407
Y1:  26386304610883
c0:  36621874029161643309794237076
c1: -4389786114023453402424
c2: -26705947173531107
c3:  316114064
c4:  240
rlim: 1800000
alim: 1800000
lpbr: 26
lpba: 26
mfbr: 52
mfba: 52
rlambda: 2.500000
alambda: 2.500000

Just as it finished running the lattice sieving I saw something about 'cat' could not be run, not sure exactly what it said because it was only there for a moment before it crashed and disappeared. And it didn't do any of the poly testing, just the lattice sieve from 900000 to 940000.

[bsquared]

Originally Posted by mnh001

...crashed and disappeared.

It shouldn't do that... I don't know how you are running the program but in a console window the text won't disappear. Running in a console window is the only method of running yafu that I know how to support.