This is my machinefile<div><div>node-1-16 slots=2</div><div>node-1-17 slots=2</div><div>node-1-18 slots=2</div><div>node-1-19 slots=2</div><div>node-1-20 slots=2</div><div>node-1-21 slots=2</div><div>node-1-22 slots=2</div>
<div>node-1-23 slots=2</div><div><br></div><div>Each cluster node has 2 processors. I launch my application with 3 processes, one on node-1-16 (manager) and two on node-1-17(workers). Two processes on node-1-17 communicate each other.</div>
<br><div class="gmail_quote">2011/3/27 Michele Marena <span dir="ltr">&lt;<a href="mailto:michelemarena@gmail.com">michelemarena@gmail.com</a>&gt;</span><br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
Hi,<div>My application performs good without shared memory utilization, but with shared memory I get performance worst than without of it.</div><div>Do I make a mistake? Don&#39;t I pay attention to something?</div><div>I know OpenMPI uses /tmp directory to allocate shared memory and it is in the local filesystem.</div>

<div><br></div><div>I thank you.</div><div>Michele.</div>
</blockquote></div><br></div>