[Cado-nfs-discuss] Linear Algebra with MPI

Pierpaolo Santucci santucci.pierpaolo at gmail.com
Thu Aug 14 20:00:04 CEST 2014


Thanks for the quick response.

In my local.sh there is:

MPI=/home/user/mpich-3.1.2/

Other steps of the GNFS (e.g. Polynomial Selection or Sieving) use MPI
correctly.

Regards,

Pierpaolo Santucci



On 14 August 2014 19:47, Emmanuel Thomé <emmanuel.thome at gmail.com> wrote:

> Most probably you haven't compiled cado-nfs for mpi use. There is an
> influential MPI variable which can be set from local.sh .
>
> Please report how far you can go. I have little to no internet connection
> at all at the moment, but I'll try to help to the extent possible.
>
> E.
> Le 14 août 2014 19:20, "Pierpaolo Santucci" <santucci.pierpaolo at gmail.com>
> a écrit :
>
>>  Hello everyone!
>>
>> I'm interested in CADO-NFS 2.1 linear algebra. I want to execute the
>> Block Wiedemann Algorithm on several hosts via MPI (mpich-3.1.2) but the
>> command:
>>
>> ./bwc.pl :complete mpi=2 hosts=host1,host2 thr=4 mn=64 nullspace=left
>> interval=100 matrix=/tmp/cado.XXXXBy837j/c90.merge.small.bin wdir=/tmp/tmp
>> interleaving=0 shuffled_product=1
>>
>> gives me the following error:
>>
>> Inconsistency -- exactly 2 == 1 * 2 MPI jobs are needed -- got 1
>> ./dispatch: exited with status 1
>>
>> What's wrong?
>>
>> Thanking you in advance.
>> Regards,
>>
>> Pierpaolo Santucci
>>
>> _______________________________________________
>> Cado-nfs-discuss mailing list
>> Cado-nfs-discuss at lists.gforge.inria.fr
>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/cado-nfs-discuss
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.gforge.inria.fr/pipermail/cado-nfs-discuss/attachments/20140814/3f211c89/attachment.html>


More information about the Cado-nfs-discuss mailing list