Discussion related to cado-nfs

[ROBERT DUCHATEAU <rjd1212@cox.net>]

Dear Paul,

 I apologize for not saving that last log file. Nevertheless, when the present running  C180 job is finished, that file will definitely be saved, including all of the other work files.

Hopefully this present factoring run will not hang endlessly during the merge phase  without entering into the Linalg stage at some point. (without eating up all available RAM in the process)

It is refreshing to hear that in your case, that C180.merge.stdout. 1file has more than enough memory to accommodate 64GB. ( RAM now upgraded to 128GB on my server )

Best Regards,

RJD

> On March 7, 2018 at 2:55 AM paul zimmermann <Paul.Zimmermann@inria.fr> wrote:
> 
> 
>        Dear Robert,
> 
> about issue 1), there is not much we can do without the log file.
> I am currently factoring a c180 with git revision 80885c6, and the
> c180.merge.stdout.1 file ends with:
> 
> Final matrix has N=28030179 nc=28030019 (160) W=4765130522 W*N=1.34e+17 W/N=170.00
> Total merge time: 39073.14 seconds
> Total usage: time 39073s (cpu), 35270s (wct) ; memory 21240M, peak 42209M
> 
> thus it should fit into 64GB.
> 
> About issue 2), this is normal. What happens is that the bkmult parameter is
> dynamically adjusted if too small. Maybe the warning message is too verbose.
> 
> Best regards,
> Paul Zimmermann