<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 10pt;
font-family:Verdana
}
--></style>
</head>
<body class='hmmessage'>
Ah, I noticed just now that the Xeon model is in the subject line.<br>So it is not a Nehalem Xeon. Then the difference is somewhat larger than I expected.<br>Maybe using fortran could help a bit.<br><br>Berk<br><br><hr id="stopSpelling">From: gmx3@hotmail.com<br>To: gmx-users@gromacs.org<br>Subject: RE: [gmx-users] Power6 @ 4.7 Ghz vs Xeon E5472 @ 3.00GHz<br>Date: Wed, 17 Feb 2010 11:47:38 +0100<br><br>
<style>
.ExternalClass .ecxhmmessage P
{padding:0px;}
.ExternalClass body.ecxhmmessage
{font-size:10pt;font-family:Verdana;}
</style>
<br><br>> Date: Wed, 17 Feb 2010 03:12:57 +1100<br>> From: Mark.Abraham@anu.edu.au<br>> To: gmx-users@gromacs.org<br>> Subject: Re: [gmx-users] Power6 @ 4.7 Ghz vs Xeon E5472 @ 3.00GHz<br>> <br>> On 17/02/10 03:06, pim schravendijk wrote:<br>> > Dear Gromacs users,<br>> ><br>> > I have been installing Gromacs on a Power6 4.7 GHz with Infiniband,<br>> > GBit-Ethernet<br>> ><br>> > before calling configure, I set:<br>> ><br>> > export LDFLAGS="-L/gpfs/schraven/fftw3/lib"<br>> > export CPPFLAGS="-I/gpfs/schraven/fftw3/include"<br>> > export CC="mpcc_r -q64"<br>> > export CFLAGS="-O3 -qstrict -qarch=auto -qtune=auto"<br>> > export CXX="mpCC_r -q64"<br>> > export CXXFLAGS="-O3 -qstrict -qarch=auto -qtune=auto"<br>> > export F77="mpxlf_r -q64"<br>> > ### O3 gives warning: The NOSTRICT option (default at OPT(3)) has the<br>> > potential to alter the semantics of a program.<br>> > export FFLAGS="-O3 -qstrict -qnoprefetch -qarch=auto -qtune=auto"<br>> > export LIBS="-lmass"<br>> > export FLIBS= "-xlf -xlf90"<br>> > export OBJECT_MODE=64<br>> ><br>> > I run on both machines at 8 cores (so within the board), and find the 8<br>> > power6 cores to be 4-5 times slower (!) than the 8 intel xeon cores.<br>> ><br>> > Does anyone have similar experience between these machines? Are there<br>> > some things I should check?<br>> > It's a bit useless to use the power6 at all with these kind of speeds,<br>> > but they were already here anyway and amazon's 30 day refund on retour<br>> > has expired already ;)<br>> > Hopefully I'm doing something wrong and the situation can still be improved?<br>> <br>> This issue might have been discussed on the list before, have a search. <br>> I suspect the issue is that there are no optimized inner loops for <br>> Power6 as there are for x86. You can see what GROMACS is finding and <br>> doing in the configure output. Probably GROMACS is finding nothing <br>> useful and falling back on generic C versions of such loops. You should <br>> experiment with --enable-fortran to try the generic FORTRAN versions.<br><br>This is partly the issue.<br>The other problem is, if I recall correctly, that someone at IBM told me<br>that the VMX units have been put in the power6 at a late state in the design process<br>and that they are therefore not very fast.<br><br>Is the factor 4 compared to a Nehalem Xeon?<br>That is roughly the difference I would expect.<br><br>Berk<br><br>> <br>> Mark<br>> -- <br>> gmx-users mailing list gmx-users@gromacs.org<br>> http://lists.gromacs.org/mailman/listinfo/gmx-users<br>> Please search the archive at http://www.gromacs.org/search before posting!<br>> Please don't post (un)subscribe requests to the list. Use the <br>> www interface or send it to gmx-users-request@gromacs.org.<br>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php<br> <br><hr>New Windows 7: Find the right PC for you. <a href="http://windows.microsoft.com/shop">Learn more.</a> <br /><hr />Express yourself instantly with MSN Messenger! <a href='http://clk.atdmt.com/AVE/go/onm00200471ave/direct/01/' target='_new'>MSN Messenger</a></body>
</html>