Several values for the number of relations needed for GNFS tasks >= 158:
* C160_133_109: 64.7M relations, 3 x 400K + 219108 cliques removed => only slight oversieving.
* C160_131_87: 62.2M relations, 2 x 400K + 381234 cliques removed => only slight oversieving.
* C163_142_104: 72.6M relations: filtering stops at singleton removal => undersieving. I expected undersieving, because:
* Aliquot 4788 index 2509 (C163): 104.5M relations, 2 x 400K + 227548 cliques removed => only slight oversieving ( [url]http://www.mersenneforum.org/showpost.php?p=203155&postcount=126[/url] )

Coarse upper bound:
* C158_136_107: 70.7M relations, 12 x 400K + 259539 cliques removed => oversieving.

Once in a while the siever terminates with the following error[CODE]appending spairs.out.T0 to spairs.out
appending spairs.out.T1 to spairs.out
-> Return value -1073741819. Updating job file and terminating...
appending spairs.out to spairs.add
Terminating...[/CODE]I work around this by increasing the special Q in the job (or resume) file, but no work is done until I notice this.

Would it be possible for the script to catch this error, increase q0 by a couple of K and resume sieving?

[QUOTE=smh;205918]Once in a while the siever terminates with the following error[CODE]appending spairs.out.T0 to spairs.out
appending spairs.out.T1 to spairs.out
-> Return value -1073741819. Updating job file and terminating...
appending spairs.out to spairs.add
Terminating...[/CODE]I work around this by increasing the special Q in the job (or resume) file, but no work is done until I notice this.

Would it be possible for the script to catch this error, increase q0 by a couple of K and resume sieving?[/QUOTE]

Could this error be the reason that the siever (in windows, especially 11e and 12e) crashes once in a while with a windows error message? Could it be a hidden bug in the siever, possibly a segfault or another kind of memory exception?

[quote=Andi47;205922]Could this error be the reason that the siever (in windows, especially 11e and 12e) crashes once in a while with a windows error message? Could it be a hidden bug in the siever, possibly a segfault or another kind of memory exception?[/quote]

You could well be right since the Python code has detected an error return from one of the siever threads.

Brian

I thought that I should let users of my Python driver know that I have slightly changed its behaviour when running the GPU version of msieve. It used to be necessary to put msieve's *.ptx files in the current working directory.

But I haave now changed this in the current version so that these files don't need to be moved. I have also added a switch to set the GPU number that msieve will use.

Brian

[QUOTE=Brian Gladman;205956]You could well be right since the Python code has detected an error return from one of the siever threads.

Brian[/QUOTE]Will you be able to let the script catch this and continue sieving at a larger q?

[quote=smh;206262]Will you be able to let the script catch this and continue sieving at a larger q?[/quote]

I will look at this but it in a part of the code that was hard to get right so I am reluctant to fiddle with it without careful thought.

Is it clear from the siever code what that error code means?

Brian

polynomial missing or corrupt
 

Hi, I've successfully used this driver on my system to factor the example from this guide: [URL]http://gilchrist.ca/jeff/factoring/nfs_beginners_guide.html[/URL]. Based on this result I attempted to factor a larger number (154 digits) using the same procedure. After approximately 10 days of polynomial selection, I observed the following prompt:

[I]polynomial selection complete
error: polynomial is missing or corrupt
Msieve Error: return value -1. Terminating...[/I]

I noticed the files, '[I]example.fb[/I]' and '[I]example.job.resume[/I]', in my example folder so I attempted to run the driver again. This returned the following:

[I]c:\ggnfs\example>..\factmsieve.py example
-> ________________________________________________________________
-> | Running factmsieve.py, a Python driver for MSIEVE with GGNFS |
-> | sieving support. It is Copyright, 2010, Brian Gladman and is |
-> | a conversion of factmsieve.pl that is Copyright, 2004, Chris |
-> | Monico. This is version 0.52, dated 9th February 2010. |
-> |______________________________________________________________|
-> This is client 1 of 1
-> Using 8 threads
-> Working with NAME = example
-> Converting msieve polynomial (*.fb) to ggnfs (*.poly) format.
Traceback (most recent call last):
File "C:\ggnfs\factmsieve.py", line 1935, in <module>
fb_to_poly()
File "C:\ggnfs\factmsieve.py", line 877, in fb_to_poly
print('skew: {0:<10.2f}'.format(skew), file = out_f)
UnboundLocalError: local variable 'skew' referenced before assignment[/I]

I'm not sure where to go from here, will I need to start the polynomial selection process over again? Is there additional configuration I should perform to ensure a successful factorization attempt? Is there a way to recover the results from this failed attempt and continue the factorization process?

Any help is greatly appreciated, thanks!
-Yogesh

If you could show us the contents of your 'example.fb' file, it may be possible to see what might be going wrong.

Brian

Thanks for the reply!

Here is the contents of example.fb:
[I]
N 9312646228577406268788416442384901501202477191719973059983726869756209036259130468605151122130796299189916412100984591030888717904364529546979463145365927
R0 0
A0 0[/I]

-Yogesh

It looks like the polynomial selection went wrong in some way. Did the script produce an example.log file - If so can you upload it here as a zip attachment?

Brian