<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <p>x lapw1  -nmat_only</p>
    <p><br>
    </p>
    <p>can be run quickly after   init_lapw. It runs in fractions of
      seconds on the frontend and just determines the matrix size and
      prints it into   case.nmat_only .</p>
    <p><br>
    </p>
    <p>From the matrix size and the knowledge if you have inversion or
      not (case.in1 or in1c) you can estimate the total memory, and also
      estimate how many cores you should use.</p>
    <p><br>
    </p>
    <p>You have  NMAT~84000 and no inversion. Memory: 10^5 * 10^5 *16  ~
      10^11 bytes ~ 100 GB for one matrix.</p>
    <p><br>
    </p>
    <p>One needs approximately 3-4 such matrices, so I would request so
      many nodes that you have  at least 400 GB available. Usually you
      don't have this on one node.In addition, one node is much too slow
      for this. <br>
    </p>
    <p>As I said before, you want sub-matrices bigger than ~2000, but
      also not larger than something like 8000.</p>
    <p>Thus you need at least   10x10=100 cores (if they have enough
      memory; and adapt the number to a multiple of your cores/node (36
      * 3 ??), but much better 20x20 or even 30x30 cores.  <br>
    </p>
    <p><br>
    </p>
    <p><br>
    </p>
    <div class="moz-cite-prefix">Am 04.07.2023 um 17:01 schrieb Ilias,
      Miroslav:<br>
    </div>
    <blockquote type="cite"
      cite="mid:5e098c61fedc494a8194df656e694b65@gsi.de">
      <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
      <style type="text/css" style="display:none;">P {margin-top:0;margin-bottom:0;}</style>
      <div id="divtagdefaultwrapper"
style="font-size:12pt;color:#000000;font-family:Calibri,Helvetica,sans-serif;"
        dir="ltr">
        <p>Greetings,</p>
        <p><br>
        </p>
        <p>I have the big system of <a
href="https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg22609.html"
            class="OWAAutoLink moz-txt-link-freetext"
            moz-do-not-send="true">
https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg22609.html</a>
          .</p>
        <p><br>
        </p>
        <p>After fixing the proper openmpi compilation of wien2k I
          proceed further into the lapw1_mpi module. But here I got the
          error "<span style="font-family:monospace"><span
              style="color:#000000;background-color:#ffffff;">xxmr2d:out
              of memory</span></span>" for SBATCH parameters  N=1,  n=36
          (that is 36 MPI threads), and <span
            style="font-family:monospace">
            <span style="color:#000000;background-color:#ffffff;">--mem=450GB</span>.<br>
          </span></p>
        <p><br>
        </p>
        <p>Now I am looking for a method for estimating the total memory
          consumption, --mem.</p>
        <p><br>
        </p>
        <p>In the  <span style="font-family:monospace"><span
              style="font-family:monospace"><span
                style="color:#ffffff;background-color:#000000;">TsOOQg.output1_1</span>
              file I see some info about memory spending in MB...Would
              it be please possible to estimate the SBATCH memory need
              from this info ?</span></span></p>
        <p><span style="font-family:monospace"><span
              style="font-family:monospace"><br>
            </span></span></p>
        <p><span style="font-family:monospace"><span
              style="font-family:monospace"><br>
            </span></span></p>
        <p><span style="font-family:monospace"><span
              style="font-family:monospace"><br>
            </span></span></p>
        <p><span style="font-family:monospace"><span
              style="font-family:monospace"><span
                style="font-family:monospace"><span
                  style="color:#000000;background-color:#ffffff;">MPI-parallel
                  calculation using    36 processors
                </span><br>
                Scalapack processors array (row,col):   6   6 <br>
                Matrix size        84052 <br>
                Optimum Blocksize for setup 124 Excess %  0.428D-01 <br>
                Optimum Blocksize for diag  30 Excess %  0.143D-01 <br>
                Base Blocksize   64 Diagonalization   32 <br>
                         allocate H      2997.6 MB          dimensions
                14016 14016 <br>
                         allocate S      2997.6 MB          dimensions
                14016 14016 <br>
                    allocate spanel        13.7 MB          dimensions
                14016    64 <br>
                    allocate hpanel        13.7 MB          dimensions
                14016    64 <br>
                  allocate spanelus        13.7 MB          dimensions
                14016    64 <br>
                      allocate slen         6.8 MB          dimensions
                14016    64 <br>
                        allocate x2         6.8 MB          dimensions
                14016    64 <br>
                  allocate legendre        89.0 MB          dimensions
                14016    13    64 <br>
                allocate al,bl (row)         4.7 MB          dimensions
                14016    11 <br>
                allocate al,bl (col)         0.0 MB          dimensions
                   64    11 <br>
                        allocate YL         3.4 MB          dimensions
                   15 14016     1 <br>
                number of local orbitals, nlo (hamilt)     1401 <br>
                      allocate YL          20.5 MB          dimensions
                   15 84052     1 <br>
                      allocate phsc         1.3 MB          dimensions
                84052</span></span></span></p>
        <p><span style="font-family:monospace"><span
              style="font-family:monospace"><span
                style="font-family:monospace"><span
                  style="font-family:monospace"><span
                    style="color:#000000;background-color:#ffffff;">Time
                    for al,bl    (hamilt, cpu/wall) :        24.49
                          24.82
                  </span><br>
                  Time for legendre (hamilt, cpu/wall) :         5.19
                         5.21<br>
                  <br>
                  <br>
                </span>And <span style="font-family:monospace"><span
                    style="color:#ffffff;background-color:#000000;">lapw1.error 
                  </span>:   <br>
                </span><br>
              </span><span style="font-family:monospace"><span
                  style="color:#000000;background-color:#ffffff;">**
                   Error in Parallel LAPW1
                </span><br>
                **  LAPW1 STOPPED at Tue 04 Jul 2023 10:36:17 AM CEST <br>
                **  check ERROR FILES! <br>
                Error in LAPW1<br>
                <br>
              </span>Best,</span></span></p>
        <p><span style="font-family:monospace"><span
              style="font-family:monospace"><br>
            </span></span></p>
        <p><span style="font-family:monospace"><span
              style="font-family:monospace">Miro<br>
            </span> <br>
          </span><br>
        </p>
        <p><br>
        </p>
        <p><br>
        </p>
        <p><br>
        </p>
        <p><br>
        </p>
      </div>
      <br>
      <fieldset class="moz-mime-attachment-header"></fieldset>
      <pre class="moz-quote-pre" wrap="">_______________________________________________
Wien mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Wien@zeus.theochem.tuwien.ac.at">Wien@zeus.theochem.tuwien.ac.at</a>
<a class="moz-txt-link-freetext" href="http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien">http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien</a>
SEARCH the MAILING-LIST at:  <a class="moz-txt-link-freetext" href="http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html">http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html</a>
</pre>
    </blockquote>
    <pre class="moz-signature" cols="72">-- 
-----------------------------------------------------------------------
Peter Blaha,  Inst. f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-158801165300
Email: <a class="moz-txt-link-abbreviated" href="mailto:peter.blaha@tuwien.ac.at">peter.blaha@tuwien.ac.at</a>          
WWW:   <a class="moz-txt-link-freetext" href="http://www.imc.tuwien.ac.at">http://www.imc.tuwien.ac.at</a>      WIEN2k: <a class="moz-txt-link-freetext" href="http://www.wien2k.at">http://www.wien2k.at</a>
-------------------------------------------------------------------------</pre>
  </body>
</html>