|
2013-12-11, 21:58
|
#727 |
"Ed Hall"
Dec 2009
Adirondack Mtns
23·167 Posts |
Additional relations from other machines
Not sure this post/query belongs here, but I'll start here, for now, since it involves factmsieve.py.
First, I have many different scripts that sieve across multiple machines using the gnfs sievers and they all seem to work fine for the most part. However, I have recently set up a new set of scripts to work with Aliqueit across several machines, but am experiencing "error -11 reading relation ######" for what seems to be all of my externally sieved relations. I'll describe my procedure, so that I can be told how ignorant I am and maybe among the details someone can decrease my ignorance. When Aliqueit finishes its battery of ECM routines, which I handle via an intermediate script (different subject), and initiates factmsieve.py, I have a script that catches the fact that gnfs_<number> has been created. This script stands by for the first presence of test.job.T0 and uses it to create working polys (named numpoly) for the other machines. Effectively, I simply reuse everything except the q0 value, which I control via the script. The script then assigns the new versions of numpoly to external machines which use the following command to initiate gnfs: Code:
gnfs-lasieve4I##e -k -a -v -o spairsM<q0 value> numpoly I have noticed the addition of -n0/n1/n2/n3 to the command lines on the original machine, but can find no reference to their use in the documentation. For that matter, I haven't been able to find a description for the -k option, either. (My search has only been within the documentation in the ggnfs files.) Could the lack of the -n# option on my external machines be the source of my trouble? Any thoughts/suggestions are welcome. Thanks to all. P.S. @Brian: I never did get mpi working within the script, but have put that on a back burner for now. The current script (above) has no modifications from your original 0.76. |
|
|
|
2013-12-11, 22:33
|
#728 |
|
Just call me Henry
"David"
Sep 2007
Cambridge (GMT/BST)
133778 Posts |
-11 means that after dividing by all the factors in the dat file and trial dividing upto 1000 the result isn't 1 for the rational polynomial.
Looking at the source for gnfs-lasieve Code:
while((option= getopt(argc,argv,"FJ:L:M:N:P:RS:ab:c:f:i:kn:o:q:rt:vz"))!=-1) {
switch(option) {
case 'R':
g_resume = 1; break;
case'F':
force_aFBcalc= 1;
break;
case'J':
NumRead(J_bits);
break;
case'L':
sysload_cmd= optarg;
break;
case'M':
if(sscanf(optarg,"%hu",&cmdline_first_mpqs_side)!=1)
complain("-M %s ???\n",optarg);
break;
case'P':
if(sscanf(optarg,"%hu",&cmdline_first_psp_side)!=1)
complain("-P %s ???\n",optarg);
break;
case'S':
if(sscanf(optarg,"%f",&sigma)!=1) {
errprintf("Cannot read floating point number %s\n",optarg);
Usage();
}
break;
case'a':
if(special_q_side!=NO_SIDE) {
errprintf("Ignoring -a\n");
break;
}
special_q_side= ALGEBRAIC_SIDE;
break;
case'b':
if(base_name!=NULL)errprintf("Ignoring -b %s\n",base_name);
else base_name= optarg;
break;
case'c':
NumRead(sieve_count);
break;
case'f':
if(sscanf(optarg,"%u:%u:%u",&first_spq,&first_spq1,&first_root)!=3) {
if(sscanf(optarg,"%u",&first_spq) == 1) {
first_spq1= first_spq;
first_root= 0;
} else Usage();
} else append_output= 1;
break;
case'i':
if(sscanf(optarg,"%hu",&cmdline_first_sieve_side)!=1)
complain("-i %s ???\n",optarg);
break;
case'k':
keep_factorbase= 1;
break;
case'n':
catch_signals= 1;
case'N':
NumRead(process_no);
break;
case'o':
g_ofile_name= optarg;
break;
case'q':
NumRead16(special_q_side);
break;
case'r':
if(special_q_side!=NO_SIDE) {
errprintf("Ignoring -r\n");
break;
}
special_q_side= RATIONAL_SIDE;
break;
case't':
if(sscanf(optarg,"%hu",&cmdline_first_td_side)!=1)
complain("-t %s ???\n",optarg);
break;
case'v':
verbose++;
break;
case'z':
zip_output= 1;
break;
}
}
Last fiddled with by henryzz on 2013-12-11 at 22:38 |
|
|
|
2013-12-11, 23:20
|
#729 | |
|
"Ed Hall"
Dec 2009
Adirondack Mtns
1111000000012 Posts |
Quote:
 As for the error, I guess I'm still in the dark. I'm sieving the algebraic side and this error is showing for every relation within a huge range: Code:
Tue Dec 10 22:52:43 2013 error -11 reading relation 520432 Tue Dec 10 22:52:43 2013 error -11 reading relation 520432 Tue Dec 10 22:52:43 2013 error -11 reading relation 520432 ... Tue Dec 10 22:53:10 2013 error -11 reading relation 1494239 Tue Dec 10 22:53:10 2013 error -11 reading relation 1494240 Tue Dec 10 22:53:10 2013 error -11 reading relation 1494241 
|
|
|
|
|
|
2013-12-11, 23:26
|
#730 |
|
"Ed Hall"
Dec 2009
Adirondack Mtns
74018 Posts |
Additional trouble
Now I'm even getting new problems!
host machine: Code:
-> gnfs-lasieve4I12e -k -o spairs.out.T0 -v -n0 -a test.job.T0 Code:
gnfs-lasieve4I12e -k -a -v -o spairsM<q0 value> numpoly Code:
... gnfs-lasieve4I12e: L1_BITS=15, SVN $Revision$ FBsize 96624+0 (deg 4), 183071+0 (deg 1) xFBgen: 1 not a root mod 3 Code:
n: 234116943738789379580780771281706880155166789502714888373927832920971823160511569007205692351556999243201 c4: 696 c3: 1877680604 c2: -6357206458268583 c1: 42147300617135555669784 c0: -81999997347928637970881448660 Y1: 9662273440429 Y0: -24082733738201157025854559 skew: 4717037.54 rlim: 2500000 alim: 1249999 lpbr: 26 lpba: 26 mfbr: 49 mfba: 49 rlambda: 2.6 alambda: 2.6 q0: 1250000 qintsize: 37500 #q1:1287500 Code:
n: 234116943738789379580780771281706880155166789502714888373927832920971823160511569007205692351556999243201 c4: 696 c3: 1877680604 c2: -6357206458268583 c1: 42147300617135555669784 c0: -81999997347928637970881448660 Y1: 9662273440429 Y0: -24082733738201157025854559 skew: 4717037.54 rlim: 2500000 alim: 1249999 lpbr: 26 lpba: 26 mfbr: 49 mfba: 49 rlambda: 2.6 alambda: 2.6 q0: 2500000 qintsize: 37500 #q1:1287500 Thanks for any and all assistance... |
|
|
|
|
2013-12-12, 14:23
|
#731 |
|
"Ed Hall"
Dec 2009
Adirondack Mtns
23·167 Posts |
OK, the new problem is fixed! I had a conflicting numpoly.afb.0 file. Maybe this was the trouble with the errors, too?
|
|
|
|
|
2013-12-12, 21:57
|
#732 |
|
"Ed Hall"
Dec 2009
Adirondack Mtns
23×167 Posts |
Ok, all appears to be working correctly now. Apparently, the problem was that I was using the same name (numpoly) for more than one instance of the siever and all the instances were writing to a single numpoly.afb.0 file which was not handling it well. I have changed the scripts to use unique numpolys and all seems well.
|
|
|
|
|
2013-12-26, 17:10
|
#733 |
|
"Ed Hall"
Dec 2009
Adirondack Mtns
23×167 Posts |
MPI Modifications to Script Are Successful
@Brian: Is this a typo in line 1060 of the original version 0.76?:
Code:
print('{0:s}, {1:d}'.format(tu[1], int(tu[0]) - int(tu[1])), FILE = outf)
 Here is a link to the current file: factmsievempi.py All comments/suggestions welcome... |
|
|
|
|
2014-04-22, 16:25
|
#734 |
|
I moo ablest echo power!
May 2013
33658 Posts |
I'm having some trouble with factmsieve on Ubuntu, and I think it's a stupid error, but I don't know why I'm suddenly getting it. I'm trying to factor a simple C122. I was able to use factmsieve on a c151 previously (on the exact same system) without issue. Now, however, it fails saying that readPoly() can't find a valid N.
Both the C122 and C151 fb files were created on a Windows machine, but as I said, the C151 worked fine previously. Any ideas on what I'm missing here? |
|
|
|
|
2014-04-22, 18:07
|
#735 |
Sep 2009
11110100012 Posts |
Try changing the line endings to LF for the Linux machine.
|
|
|
|
|
2014-04-22, 19:53
|
#736 |
|
I moo ablest echo power!
May 2013
13×137 Posts |
Tried that (using Notepad++), and I get the same error. It's failing when it calls procrels. I had to compile that (and the other non-lasieve_4 64-bit files) myself since I only had the Windows binaries. Could that have something to do with it?
Edit: Seems the issue was with the .ini file. I used echo to output the number to a new ini in Terminal, and now everything is hunky-dory. Thanks for the help! Last fiddled with by wombatman on 2014-04-22 at 20:21 |
|
|
|
|
2014-05-30, 17:07
|
#737 |
|
I moo ablest echo power!
May 2013
13·137 Posts |
I'm still having the same issue as before, now with the fb file for a C122. I've copied the file onto my Ubuntu computer, opened it in GEdit, and changed the line endings to Unix/Linux. I did the same echo to file for the c122.ini.
And I still get the Code:
readPoly(): Did not find a valid 'n' value! |
|
|
|
 |
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| Msieve & ggnfs on MacOS | xilman | Msieve | 8 | 2017-05-20 00:12 |
| Factorizing with MSIEVE, GGNFS & Factmsieve.py | Romuald | Msieve | 24 | 2015-11-09 20:16 |
| Infinite loop for ggnfs or msieve | Greebley | Aliquot Sequences | 4 | 2013-02-06 19:28 |
| Error running GGNFS+msieve+factmsieve.py | D. B. Staple | Factoring | 6 | 2011-06-12 22:23 |
| A new driver? (or type of driver?) | 10metreh | Aliquot Sequences | 3 | 2010-02-15 15:57 |
