<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 10pt;
font-family:Tahoma
}
--></style>
</head>
<body class='hmmessage'>
Hi,<br><br>I realized now that this is an SSE issue.<br>Normally you would get NAN (or is it INF?). That is treated correctly in the GROMACS TPI code.<br>But in SSE a float "wraps around" when it overflows, which could, in very few cases, lead to a reasonably<br>looking energy value (I check for very high and very low values).<br>I found that you can check for overflows in SSE and committed a fix for 4.5.2.<br>I also filled the first 10 points (up to r=0.02 nm) of the potential/force tables, these used to be zero.<br>These values are only relevant for energy minimization or TPI with extreme atomic overlap.<br><br>Berk<br><br><hr id="stopSpelling">From: gmx3@hotmail.com<br>To: gmx-users@gromacs.org<br>Subject: RE: [gmx-users] Overflow problem with test-particle insertion<br>Date: Thu, 9 Sep 2010 09:39:42 +0200<br><br>
<meta http-equiv="Content-Type" content="text/html; charset=unicode">
<meta name="Generator" content="Microsoft SafeHTML">
<style>
.ExternalClass .ecxhmmessage P
{padding:0px;}
.ExternalClass body.ecxhmmessage
{font-size:10pt;font-family:Tahoma;}
</style>
Hi,<br><br>This is an interesting issue.<br>The chance is quite small that this happens, but maybe not negligible.<br>In single precision the maximum a float can store is 2^127.<br>This gives a minimum distance of (2^127)^-1/12 = 6.5e-4 nm.<br>The chance of inserting a particle within this radius is dens*3e-10,<br>where dens is the number of particles per nm^3.<br>A typical density of LJ particles is 30 per nm^3, which leads to a chance of 1e-8.<br>Such insertion numbers can be reached, so we probably have to worry about this.<br><br>However, in your example the distance seems to be around 4e-3, which would<br>give r^-12 = 6e28. This still fits in a float and should not cause problems.<br>So we should make sure we understand what's going on here.<br>Could you file a bugzilla with the files to reproduce this and which insertion<br>is the problematic one?<br><br>I so two possible solutions:<br>Force tabulated potentials with TPI, this can currently be achieved by setting<br>the environment variable GMX_FORCE_TABLES<br>Or require double precision.<br>But I think both solutions would lead to about 40% lower performance.<br><br>Berk<br><br><br><hr id="ecxstopSpelling">Date: Wed, 8 Sep 2010 21:16:46 -0400<br>From: kdaly@princeton.edu<br>To: gmx-users@gromacs.org<br>Subject: [gmx-users] Overflow problem with test-particle insertion<br><br>Hello Gromacs users,<br><br>I sent a message to the list in June describing what appeared to be a float overflow issue with the energy calculation for test-particle insertions: <a href="http://lists.gromacs.org/pipermail/gmx-users/2010-June/052213.html" target="_blank">http://lists.gromacs.org/pipermail/gmx-users/2010-June/052213.html</a>. <br>
<br>I have recently tried the test-particle insertion mode in Gromacs-4.5.1, and it seems the problem is still there. Does anyone know how to work around or fix this problem without using tabulated potentials?<br><br>-Kevin<br>
<br>--
gmx-users mailing list gmx-users@gromacs.org
http://lists.gromacs.org/mailman/listinfo/gmx-users
Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
Please don't post (un)subscribe requests to the list. Use the
www interface or send it to gmx-users-request@gromacs.org.
Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
<br>--
gmx-users mailing list gmx-users@gromacs.org
http://lists.gromacs.org/mailman/listinfo/gmx-users
Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
Please don't post (un)subscribe requests to the list. Use the
www interface or send it to gmx-users-request@gromacs.org.
Can't post? Read http://www.gromacs.org/Support/Mailing_Lists </body>
</html>