Hi Ralph,
thanks a lot, including it in the next release would be great.
Best regards
Igor
On Wed, 30 Nov 2011 14:30:25 -0700
Ralph Castain wrote:
> Hi Igor
>
> As I recall, this eventually traced back to a change in slurm at some point.
> I believe the latest interpretatio
lots_max = 0;
-node->slots = slots[i] / cpus_per_task;
+/* Don't divide by cpus_per_task */
+node->slots = slots[i];
opal_list_append(nodelist, &node->super);
}
free(slots);
Are there situations where this might n
all size of the cluster, add "-mca routed direct" to your
> command line. This will tell mpirun to talk directly to each daemon. However,
> note that your job may still fail as the procs won't be able to open sockets
> to their peers to send MPI messages, if you use
LD_LIBRARY_PATH :) Adding
"/usr/lib/openmpi/lib" to the otherwise empty LD_LIBRARY_PATH produces
same results.
Can someone suggest a possible solution or at least a direction in
which I should continue my troubleshooting?
--
Thank you all for your time,
Igor
similarly for other collectives.
Best,
Igor
2009/7/23 Gus Correa :
> Dear OpenMPI experts
>
> I would like to experiment with the OpenMPI tuned collectives,
> hoping to improve the performance of some programs we run
> in production mode.
>
> However, I could not find any do
Hi Sangamesh,
I'd look into making sure that the node you are using is not running
anything in parallel.
Make sure you allocate a whole node and it is clean from previous jobs.
Best,
INK
Hi Sangamesh,
As far as I can tell there should be no difference if you run CPMD on a
single node whether with or without ib. One easy thing that you could do is
to repeat your runs on the infiniband node(s) with and without infiniband
using --mca btl ^tcp and --mca btl ^openib respectively. But si
procs * 8 procs/node = 8 GB/node plus you need to double
because of buffering. I was told by Mellanox (our cards are ConnectX cards)
that they introduced XRC in OFED 1.3 in addition to Share Receive Queue
which should reduce memory foot print but I have not tested this yet.
HTH,
Igor
2009/1/23 Gab
what is your message size and the number of cores per node?
is there any difference using different algorithms?
2009/1/23 Gabriele Fatigati
> Hi Jeff,
> i would like to understand why, if i run over 512 procs or more, my
> code stops over mpi collective, also with little send buffer. All
> proce
> - per the "sm" thread, you might want to try with just IB (and not
> shared memory), just to see if that helps (I don't expect that it
> will, but every situation is different). Try running "mpirun --mca
> btl openib ..." (vs. "--mca btl ^tcp").
Unfortunately you were right- it did not help. Sm
this task never completes
Thanks in advance. Sorry for the long post.
Igor
PS Im following the discussion on slow sm btl but not sure if this particular
problem is related or not. BTW the Open MPI build Im using is for Intel
compiler.
PPS MVAPICH and MVAPICH2 behave much better but not
IMB PingPong). Of course commercial MPI
libraries offer low latency too e.g. Scali MPI.
Best,
Igor
>
> --
>
> Dresden University of Technology
> Center for Information Services
> and High Performance Computing (ZIH)
> D-01062 Dresden
> Germany
>
> e-mail: andy.g
Hello,
When i try to install OpenMPI on Linux Suse 10.2 on AMDX2 Dual Core
processor i get the following message:
make[3]: Entering directory `/home/igor/openmpi-1.2.3/opal/libltdl'
if /bin/sh ./libtool --tag=CC --mode=compile gcc -DHAVE_CONFIG_H -I. -I.
-I. -D
LT_CONFIG_H=''
13 matches
Mail list logo
