Just a quick note - I have noticed the same behavior with LAM-MPI, but not with MPICH.<br>Jacques<br><br><div class="gmail_quote">2008/12/5 Paul Edwards <span dir="ltr"><<a href="mailto:paul.m.edwards@gmail.com">paul.m.edwards@gmail.com</a>></span><br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Hi all,<br><br>It just depends on how it is implemented with MPI. There will be lower latency if it spins in a loop waiting for a message. Here is a link to the FAQ which shows how you can stop this for OpenMPI:<br>
<br><a href="http://www.open-mpi.org/faq/?category=running#oversubscribing" target="_blank">http://www.open-mpi.org/faq/?category=running#oversubscribing</a><br>
<a href="http://www.open-mpi.org/faq/?category=running#force-aggressive-degraded" target="_blank">http://www.open-mpi.org/faq/?category=running#force-aggressive-degraded</a><br><br>Regards,<br>Paul<br><br><div class="gmail_quote">
2008/12/5 John M. Patchett <span dir="ltr"><<a href="mailto:patchett@lanl.gov" target="_blank">patchett@lanl.gov</a>></span><div><div></div><div class="Wj3C7c"><br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Hi Jens,<br>
If I recall correctly, your observation lies in the type of hardware you are using. I think some hardware allows a developer to leverage an interrupt while some requires polling for a received message. Design requirements for MPI to be fast with low latency usually (I presume) outweigh lowering heat production.<br>
<br>
I think I first noticed this issue with mvapich when infiniband was new. I don't think we had it with Myrinet 2000 or Gig-E.<br>
<br>
I also think ParaView is a normal mpi application, we just don't notice machines spinning while waiting during batch processing, we just assume they're working hard.<div><div></div><div><br>
<br>
-John.<br>
<br>
On Dec 5, 2008, at 9:42 AM, Jens wrote:<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Hi John,<br>
<br>
thanks for your answer. That makes sense. "Normal" mpi-apps are probably<br>
not written to wait for more things to do - they are simply always busy.<br>
<br>
It is just a pity that the cluster has to run 100% producing a lot of<br>
heat for nothing.<br>
<br>
So the MPI-lib will probably not change this behavior :( ? (I am using<br>
open-mpi 1.2.8)<br>
<br>
Greetings<br>
Jens<br>
<br>
<br>
John M. Patchett schrieb:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Hi Jens,<br>
Your pvserver is probably waiting on an MPI_Recv and your MPI<br>
implementation is spinning.<br>
You will note that process 0 probably isn't doing this, as the other<br>
nodes are waiting on process 0 to send.<br>
I have searched this problem all the way to the MPI developers as it's<br>
easy to replicate without paraview and the MPI guys assure me the<br>
alternatives are worse.<br>
-John.<br>
<br>
On Dec 5, 2008, at 8:42 AM, Jens wrote:<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Hi,<br>
<br>
if I run "mpirun -np 4 ./pvserver" on our cluster-node and connect from<br>
my client, this pvserver always shows 100% cpu usage - even if I do<br>
nothing at the client.<br>
<br>
Seems to me as if there is a loop waiting for the client to ask for<br>
action - but this loop is calling no wait/sleep function.<br>
<br>
Greetings<br>
Jens<br>
_______________________________________________<br>
ParaView mailing list<br>
<a href="mailto:ParaView@paraview.org" target="_blank">ParaView@paraview.org</a><br>
<a href="http://www.paraview.org/mailman/listinfo/paraview" target="_blank">http://www.paraview.org/mailman/listinfo/paraview</a><br>
</blockquote>
<br>
</blockquote></blockquote>
<br>
_______________________________________________<br>
ParaView mailing list<br>
<a href="mailto:ParaView@paraview.org" target="_blank">ParaView@paraview.org</a><br>
<a href="http://www.paraview.org/mailman/listinfo/paraview" target="_blank">http://www.paraview.org/mailman/listinfo/paraview</a><br>
</div></div></blockquote></div></div></div><br>
<br>_______________________________________________<br>
ParaView mailing list<br>
<a href="mailto:ParaView@paraview.org">ParaView@paraview.org</a><br>
<a href="http://www.paraview.org/mailman/listinfo/paraview" target="_blank">http://www.paraview.org/mailman/listinfo/paraview</a><br>
<br></blockquote></div><br>