[mpich-discuss] MPICH2 on AIX
Eric Sun
eric.sunyy at gmail.com
Sun Dec 13 16:13:22 CST 2009
For src/openpa/src/Makefile, if I don't modify, I got:
ar cru libopa.a opa_primitives.o opa_queue.o
ar: 0707-126 opa_primitives.o is not valid with the current object file
mode.
Use the -X option to specify the desired object mode.
So, the AR="ar -X64" set in the configure line didn't apply to this
directory.
For the main Makefile, if I don't modify the "INSTALL=confdb/install-sh"
line by giving the full path, it complains confdb/install-sh not found.
Regarding the network problem, I'll try your suggestions. BTW, for a single
SMP machine, gforker and hydra, which one is more suited?
Thanks!
On Sun, Dec 13, 2009 at 1:24 PM, Pavan Balaji <balaji at mcs.anl.gov> wrote:
>
> On 12/13/2009 11:44 AM, Eric Sun wrote:
> > FYI, I needed to edit src/openpa/src/Makefile to change "AR = ar" to "AR
> > = ar -X64". So, the AR="ar -X64" didn't apply to this directory. Also,
> > somehow I need to edit the main Makefile to add full path for "INSTALL =
> > confdb/install-sh". Everything else is fine.
>
> What errors were you seeing without these changes? This should work out
> of the box, AFAICT.
>
> > I'm now doing a first test using "mpiexec -l -n 4 hostname". For n=1, it
> > always works. n=2 occasionally fails. n=3 occasionally successes. n=4
> > always gives the following message:
> > mpiexec_pls01.ccni.rpi.edu <http://mpiexec_pls01.ccni.rpi.edu> (mpiexec
> > 392): no msg recvd from mpd when expecting ack of request
>
> This usually points to some problem with the networking setup. Can you
> use the mpdcheck utility to see if it can find any? Do you have any
> firewall for some of the ports?
>
> Alternatively, you can try using the Hydra process manager
> (mpiexec.hydra instead of mpiexec). Here's some documentation:
> http://wiki.mcs.anl.gov/mpich2/index.php/Using_the_Hydra_Process_Manager
>
> If there's a serious networking setup problem, both mpd and Hydra will
> give you errors, of course. So, you'll need to make sure your setup is
> fine in either case.
>
> -- Pavan
>
> --
> Pavan Balaji
> http://www.mcs.anl.gov/~balaji
> _______________________________________________
> mpich-discuss mailing list
> mpich-discuss at mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>
--
Yiyang Sun
Department of Physics, Applied Physics, and Astronomy
Rensselaer Polytechnic Institute
Troy, NY 12180, USA
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20091213/8b11bf70/attachment.htm>
More information about the mpich-discuss
mailing list