<html>
<head>
<meta http-equiv="Content-Type" content="text/html;
charset=windows-1252">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix">Hi,<br>
<br>
This is due to the recent commit 795ad1f8<br>
We'll fix it. Thanks for reporting this.<br>
<br>
Cheers,<br>
<br>
Berk<br>
<br>
On 2017-08-28 15:42, Johannes Wagner wrote:<br>
</div>
<blockquote type="cite"
cite="mid:54966712-5737-458E-9EC7-54EC9237C303@h-its.org">
<meta http-equiv="Content-Type" content="text/html;
charset=windows-1252">
Hey all,
<div class=""><br class="">
</div>
<div class="">I have an odd behaviour of how threads are divided
between OMP and MPI for gromacs 2017 git master (25.08.).</div>
<div class=""><br class="">
</div>
<div class="">Following setup: CentOS 7, 2x8core Xeon <b class="">v4</b>,
HT enabled, <b class="">1x1080</b> GPU and compiling with gcc
4.8.5, -DGMX_GPU=ON, hwloc installed</div>
<div class=""><br class="">
</div>
<div class="">If I run something CPU only with -nb cpu , it picks
1 MPI x 32 OMP threads and I get the following fatal error:</div>
<div class=""><br class="">
</div>
<div class="">
<div class=""><font class="" face="Roboto-Regular" size="1">Fatal
error:</font></div>
<div class=""><font class="" face="Roboto-Regular" size="1">Your
choice of 1 MPI rank and the use of 32 total threads leads
to the use of</font></div>
<div class=""><font class="" face="Roboto-Regular" size="1">32
OpenMP threads, whereas we expect the optimum to be with
more MPI ranks</font></div>
<div class=""><font class="" face="Roboto-Regular" size="1">with
1 to 6 OpenMP threads. If you want to run with this many
OpenMP threads,</font></div>
<div class=""><font class="" face="Roboto-Regular" size="1">specify
the -ntomp option. But we suggest to increase the number of
MPI ranks</font></div>
<div class=""><font class="" face="Roboto-Regular" size="1">(option
-ntmpi).</font></div>
</div>
<div class=""><br class="">
</div>
<div class="">If I run the same system with CPU+GPU, it picks 1
MPI x 32 OMP threads and works.</div>
<div class=""><br class="">
</div>
<div class="">If I do the exact same on a 2x8 core Xeon <b
class="">v3</b>, HT enabled, <b class="">2xTitanX</b> GPU,
also compiling with -DGMX_GPU=ON. Running this with or without
-nb cpu gives 8 MPI x 4 OMP and works. No errors.</div>
<div class=""><br class="">
</div>
<div class="">Compiling gromacs with -DGMX_GPU=OFF gives for both
machines 32 MPI x 1 OMP threads and is working.</div>
<div class=""><br class="">
</div>
<div class="">Does anyone have a clue what causes the different
behaviour on the 2 machines? Is it single vs. dual gpu? And
moreover why do I get that fatal error? And why is there a
different MPI/OMP configuration depending on compiling with or
without gpu support, but in both cases running on cpu only?</div>
<div class=""><br class="">
</div>
<div class="">I did not file a bug report on redmine, but could
also do. Any hints appreciated!</div>
<div class=""><br class="">
</div>
<div class=""><br class="">
</div>
<div class="">cheers, Johannes</div>
<br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<br>
</blockquote>
<br>
</body>
</html>