<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=us-ascii">
<META NAME="Generator" CONTENT="MS Exchange Server version 08.00.0681.000">
<TITLE>-nolocal switch not working</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/rtf format -->
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">Hi Guys,</FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri">I’</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">m having a problem with the</FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri">–</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">nolocal switch.</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri"> I want my cluster headnode, not to do any number-crunching, but just be use as an execution node. If I use the</FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri">–</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">nolocal</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri"> switch, the job runs only on 1 process, no matter how many I specify with</FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri">–</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">np.</FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri"></FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri"> </FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri"> Some details:</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">If I have the headnode (SIB) in my machines file, it get</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">’</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">s assigned process zero, and then mpirun starts cycling through the machines file line by line</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">, and allocated another</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri"> 2 processes to SIB</FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri">ONTOP</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri"> of process 0</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">:</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">>></FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">SIB:/mpich/examples sharcnet$ mpirun -np 6 -machinefile machines cpi</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">Process 0 on sib</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">Process 3 on node2</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">Process 2 on node1</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">Process 5 on node1</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">Process 1 on sib</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">Process 4 on sib</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">pi is approximately 3.1416009869231249, Error is 0.0000083333333318</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">wall clock time = 0.003049</FONT></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">If I leave SIB out of te machines file, it doesn</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">’</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">t get assigned the 2 addition processes, but still gets process 0, which isn</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">’</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">t just a dissemination process, it does real num</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">b</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">er-crunching</FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri">as part of the job (what I</FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri">don’t</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri"> want). If I use the</FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri">–</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">noloca command, I get the following output:</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"> <FONT FACE="Calibri">>></FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri"> mpirun -nolocal -np 4 -machinefile machines cpi</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">Process 0 on node1</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">pi is approximately 3.1416009869231254, Error is 0.0000083333333323</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">wall clock time = 0.000119</FONT></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">I tried running it with the</FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri">–</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">t switch to test only, and under that condition, it seems</FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri">to show me it SHOULD work fine</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">:</FONT></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"> <FONT FACE="Calibri">>></FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri">mpirun -t -nolocal -np 4 -machinefile machines cpi</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">Procgroup file:</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">node1 0 /mpich/examples/cpi</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">node2 1 /mpich/examples/cpi</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">node1 1 /mpich/examples/cpi</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">node2 1 /mpich/examples/cpi</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">ssh node1 "/mpich/examples/cpi" -p4pg "/mpich/examples/PI14147" -p4wd "/mpich/examples"</FONT></SPAN><SPAN LANG="en-us"></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">Yet from the second console clip, you can see it clearly doesn</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">’</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">t work.</FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">Any idea? I</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">’</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">ve done</FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri">a lot</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri"> of searching, and can</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">’</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">t find an answer. I am running a Mac cluster with</FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri">intel chips</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri"> and OS X 10.4, Mpich version 1.2.7p1</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">.</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri"> I found a mailing list</FONT></SPAN><SPAN LANG="en-us"> <FONT FACE="Calibri">thread</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri"> from 2004 with the exact same problem on</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri"> Sparc</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">’</FONT></SPAN><SPAN LANG="en-us"><FONT FACE="Calibri">s and SUSE (</FONT></SPAN><SPAN LANG="en-us"></SPAN><A HREF="http://www.beowulf.org/archive/2004-December/011510.html"><SPAN LANG="en-us"><U><FONT COLOR="#0000FF" FACE="Calibri">http://www.beowulf.org/archive/2004-December/011510.html</FONT></U></SPAN><SPAN LANG="en-us"></SPAN></A><SPAN LANG="en-us"><FONT FACE="Calibri">), no solution. </FONT></SPAN></P>
<P DIR=LTR><SPAN LANG="en-us"><FONT FACE="Calibri">-Milo</FONT></SPAN><SPAN LANG="en-us"></SPAN></P>
</BODY>
</HTML>