<html>
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
Am 19.01.2013 16:07, schrieb Paolo Giannozzi:
<blockquote
cite="mid:6E6D1143-2C37-44BE-8E66-E5508B2BD861@democritos.it"
type="cite">
<pre wrap="">are you sure you are re-running exactly the same calculation in the
same conditions?</pre>
</blockquote>
I checked everything yet another and found that previous version was
linked with ATLAS. <br>
It claims: per-process dynamical memory: 934.9 Mb<br>
Allocates 2.7 Gb and works ok. 5.0.1 with ATLAS also seems to work
(of course,
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
considerably slower). Sorry for the mistake.<br>
<br>
<blockquote
cite="mid:6E6D1143-2C37-44BE-8E66-E5508B2BD861@democritos.it"
type="cite">
<pre wrap="">You may further reduce the memory by setting
"diago_david_ndim=2", or
'diagonalization="cg"', as explained in the documentation, setting
"mixing_ndim" to
4 or so.
</pre>
</blockquote>
Thanks a lot for the suggestions! This example is not a limit of our
system sizes in coordination polymers calculations so I'll keep
tuning.<br>
<br>
Anton S. Lytvynenko,<br>
<br>
L.V.Pisarzhevskii Institute of Physical Chemistry of the National
Academy of Sciences of Ukraine.<br>
<br>
</body>
</html>