<!DOCTYPE html>
<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <p><br>
    </p>
    <div class="moz-cite-prefix">On 10/24/2025 9:16 AM, IOSIF PANAGIOTIS
      wrote:<br>
    </div>
    <blockquote type="cite"
cite="mid:GV1P250MB085762FC2CC965EC92BB7B92F9F1A@GV1P250MB0857.EURP250.PROD.OUTLOOK.COM">
      <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
style="font-family: Aptos, Aptos_EmbeddedFont, Aptos_MSFontService, Calibri, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"
        class="elementToProof">
        Dear Steve,<br>
        <br>
        Thank you for your reply.<br>
        <br>
        At this point, I reckon it makes sense to just use a full node
        (of 112 cores) for plain tests.<br>
        In any case, they will finish quickly enough and not consume a
        lot of my allocated time.<br>
      </div>
    </blockquote>
    Probably so. Maybe even a fraction of a node.
    <blockquote type="cite"
cite="mid:GV1P250MB085762FC2CC965EC92BB7B92F9F1A@GV1P250MB0857.EURP250.PROD.OUTLOOK.COM">
      <div
style="font-family: Aptos, Aptos_EmbeddedFont, Aptos_MSFontService, Calibri, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"
        class="elementToProof"><br>
        I will leave the performance/convergence study varying the
        number of cores for the time being. <br>
        <br>
        Regarding the --memory" option, I read the <a
title="https://lists.einsteintoolkit.org/pipermail/users/2025-October/009786.html"
href="https://lists.einsteintoolkit.org/pipermail/users/2025-October/009786.html"
          moz-do-not-send="true">
          minutes from the last ETK meeting</a>, and it seems that there
        is no obvious answer.<br>
      </div>
    </blockquote>
    <p>Yes. No one seems to know. I suspect there's some way to pass it
      along to the submitscript where it can be given to slurm, but I'd
      have to try and read through the source code to figure that out.</p>
    <p><br>
    </p>
    <p>--Steve</p>
    <blockquote type="cite"
cite="mid:GV1P250MB085762FC2CC965EC92BB7B92F9F1A@GV1P250MB0857.EURP250.PROD.OUTLOOK.COM">
      <div
style="font-family: Aptos, Aptos_EmbeddedFont, Aptos_MSFontService, Calibri, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"
        class="elementToProof"><br>
      </div>
      <div
style="font-family: Aptos, Aptos_EmbeddedFont, Aptos_MSFontService, Calibri, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"
        class="elementToProof">
        In case it is useful, I summarise my current understanding
        below:</div>
      <ul
data-editing-info="{"applyListStyleFromLevel":false,"unorderedStyleType":2}"
        style="margin-top: 0px; margin-bottom: 0px;">
        <li
style="font-family: Aptos, Aptos_EmbeddedFont, Aptos_MSFontService, Calibri, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); list-style-type: "- ";">
          <div role="presentation" class="elementToProof"><a
title="https://docs.einsteintoolkit.org/et-docs/Configuring_a_new_machine"
href="https://docs.einsteintoolkit.org/et-docs/Configuring_a_new_machine"
              moz-do-not-send="true">this wiki link</a> says that "<code>memory</code> is
            currently only used by simfactory's <code>distribute</code>
            utility script"</div>
        </li>
        <li
style="font-family: Aptos, Aptos_EmbeddedFont, Aptos_MSFontService, Calibri, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); list-style-type: "- ";">
          <div role="presentation" class="elementToProof">The <code>distribute</code> script
            seems to be a testing script only, so it does not seem to be
            relevant when we submit a job</div>
        </li>
        <li
style="font-family: Aptos, Aptos_EmbeddedFont, Aptos_MSFontService, Calibri, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); list-style-type: "- ";">
          <div role="presentation" class="elementToProof">There doesn't
            seem to be any source documenting exactly how simfactory
            handles memory</div>
        </li>
        <li
style="font-family: Aptos, Aptos_EmbeddedFont, Aptos_MSFontService, Calibri, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); list-style-type: "- ";">
          <div role="presentation" class="elementToProof">Without taking
            a look into the respective source code, my guess is that
            simfactory will try to use as much memory as needed by the
            job (but not necessarily the whole memory of the node)</div>
        </li>
        <li
style="font-family: Aptos, Aptos_EmbeddedFont, Aptos_MSFontService, Calibri, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); list-style-type: "- ";">
          <div role="presentation" class="elementToProof">In any case,
            if we request a full node, we will  get billed accordingly</div>
        </li>
        <li
style="font-family: Aptos, Aptos_EmbeddedFont, Aptos_MSFontService, Calibri, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); list-style-type: "- ";">
          <div role="presentation" class="elementToProof"><i>Only if we
              request less than a full node</i>, would the --memory
            option possibly come into play and affect the billing</div>
        </li>
      </ul>
      <div
style="font-family: Aptos, Aptos_EmbeddedFont, Aptos_MSFontService, Calibri, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"
        class="elementToProof">
        <br>
      </div>
      <div
style="font-family: Aptos, Aptos_EmbeddedFont, Aptos_MSFontService, Calibri, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"
        class="elementToProof">
        Best,<br>
        Panagiotis </div>
      <div
style="font-family: Aptos, Aptos_EmbeddedFont, Aptos_MSFontService, Calibri, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"
        class="elementToProof">
        <br>
        <br>
      </div>
      <hr style="display:inline-block;width:98%" tabindex="-1">
      <div id="divRplyFwdMsg" dir="ltr"><font face="Calibri, sans-serif"
          style="font-size:11pt" color="#000000"><b>From:</b> Users
          <a class="moz-txt-link-rfc2396E" href="mailto:users-bounces@einsteintoolkit.org"><users-bounces@einsteintoolkit.org></a> on behalf of Steven
          Brandt via Users <a class="moz-txt-link-rfc2396E" href="mailto:users@einsteintoolkit.org"><users@einsteintoolkit.org></a><br>
          <b>Sent:</b> Thursday, October 23, 2025 4:28 PM<br>
          <b>To:</b> <a class="moz-txt-link-abbreviated" href="mailto:users@einsteintoolkit.org">users@einsteintoolkit.org</a>
          <a class="moz-txt-link-rfc2396E" href="mailto:users@einsteintoolkit.org"><users@einsteintoolkit.org></a><br>
          <b>Subject:</b> Re: [Users] Inconsistency warnings:
          cores/threads mismatch [Leonardo cluster]</font>
        <div> </div>
      </div>
      <div>
        <p><br>
        </p>
        <div class="x_moz-cite-prefix">On 10/14/2025 3:55 AM, IOSIF
          PANAGIOTIS wrote:<br>
        </div>
        <blockquote type="cite">
          <style type="text/css" style="display:none">p
        {margin-top:0;
        margin-bottom:0}</style>
          <div class="x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
            Hi all,<br>
            <br>
            I am sending a reminder regarding two unanswered questions
            on the mailing list, in case someone has a suggestion.<br>
            <br>
          </div>
          <ol
data-editing-info="{"applyListStyleFromLevel":false,"orderedStyleType":3}"
            start="1" style="margin-top:0px; margin-bottom:0px">
            <li
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0); list-style-type:"1) "">
              <div role="presentation" class="x_elementToProof">Clarification
                about how SimFactory handles the "--memory" option and
                how this affects how one should navigate the cluster's
                billing policy:</div>
              <div
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0); list-style-type:"1) "">
                <div role="presentation" class="x_elementToProof"><a
                    id="LPlnk557853"
href="https://lists.einsteintoolkit.org/pipermail/users/2025-September/009761.html"
class="x_moz-txt-link-freetext moz-txt-link-freetext"
                    moz-do-not-send="true">https://lists.einsteintoolkit.org/pipermail/users/2025-September/009761.html</a></div>
              </div>
              <div
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0); list-style-type:"1) "">
                <div role="presentation" class="x_elementToProof"><br>
                </div>
              </div>
            </li>
            <li
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0); list-style-type:"2) "">
              <div role="presentation" class="x_elementToProof">Using
                'leonardo-dcgp.ini' and understanding how to properly
                request one full node:</div>
              <div
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0); list-style-type:"2) "">
                <div role="presentation" class="x_elementToProof"><a
                    id="LPlnk416851"
href="https://lists.einsteintoolkit.org/pipermail/users/2025-September/009762.html"
class="x_moz-txt-link-freetext moz-txt-link-freetext"
                    moz-do-not-send="true">https://lists.einsteintoolkit.org/pipermail/users/2025-September/009762.html</a></div>
              </div>
            </li>
          </ol>
        </blockquote>
        <p>Normally, one requests --procs equal to the number of cores
          on the node.</p>
        <p><br>
        </p>
        <p>So, imagine one has a machine with nodes that have 32 cores
          each.</p>
        <p><br>
        </p>
        <p>One could say --procs 32, and that should be an entire node.
          However, maybe you want to run with 8 threads per MPI task. In
          that case, you would say --procs 32 --num-threads 8.</p>
        <p><br>
        </p>
        <p>If you want to run on N nodes, then the number of procs would
          be 32*N, and Simfactory will figure it out.</p>
        <p><br>
        </p>
        <p>--Steve</p>
        <blockquote type="cite">
          <ol
data-editing-info="{"applyListStyleFromLevel":false,"orderedStyleType":3}"
            start="1" style="margin-top:0px; margin-bottom:0px">
            <li
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0); list-style-type:"2) "">
              <div
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0); list-style-type:"2) "">
              </div>
              <br>
            </li>
          </ol>
          <div class="x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
            <br>
          </div>
          <div class="x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
            Thanks,</div>
          <div class="x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
            Panayotis</div>
          <hr tabindex="-1" style="display:inline-block; width:98%">
          <div id="x_divRplyFwdMsg" dir="ltr"><font
              face="Calibri, sans-serif" color="#000000"
              style="font-size:11pt"><b>From:</b> Users
              <a class="x_moz-txt-link-rfc2396E"
                href="mailto:users-bounces@einsteintoolkit.org"
                moz-do-not-send="true">
                <users-bounces@einsteintoolkit.org></a> on behalf
              of IOSIF PANAGIOTIS <a class="x_moz-txt-link-rfc2396E"
                href="mailto:PANAGIOTIS.IOSIF@units.it"
                moz-do-not-send="true">
                <PANAGIOTIS.IOSIF@units.it></a><br>
              <b>Sent:</b> Monday, September 29, 2025 12:24 PM<br>
              <b>To:</b> Roland Haas <a class="x_moz-txt-link-rfc2396E"
                href="mailto:rhaas@mail.ubc.ca" moz-do-not-send="true">
                <rhaas@mail.ubc.ca></a>; Bruno Giacomazzo <a
                class="x_moz-txt-link-rfc2396E"
                href="mailto:bruno.giacomazzo@unimib.it"
                moz-do-not-send="true">
                <bruno.giacomazzo@unimib.it></a><br>
              <b>Cc:</b> Einstein Toolkit Users <a
                class="x_moz-txt-link-rfc2396E"
                href="mailto:users@einsteintoolkit.org"
                moz-do-not-send="true">
                <users@einsteintoolkit.org></a><br>
              <b>Subject:</b> Re: [Users] Inconsistency warnings:
              cores/threads mismatch [Leonardo cluster]</font>
            <div> </div>
          </div>
          <style type="text/css" style="display:none">p
        {margin-top:0;
        margin-bottom:0}</style>
          <div dir="ltr">
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              Hi Roland,<br>
              <br>
              Thanks for your reply.<br>
              You touch on an important point, i.e the <b>cluster's</b>
              <b>billing policy</b>, that hadn't crossed my mind.<br>
              <br>
            </div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              From the billing policy of Leonardo, it seems that <b>it
                is</b> <b>possible to use only a fraction of a node's
                total CPUs.</b></div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              <a
class="x_x_OWAAutoLink x_moz-txt-link-freetext moz-txt-link-freetext"
                id="LPlnk971574"
href="https://docs.hpc.cineca.it/hpc/hpc_intro.html#billing-policy"
                moz-do-not-send="true">https://docs.hpc.cineca.it/hpc/hpc_intro.html#billing-policy</a></div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              <br>
            </div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              <b>However</b>, the documentation also stresses that:</div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              <i>...if a job reserves all of a node’s RAM — even without
                utilizing all its CPUs — the node becomes unusable for
                other jobs and is therefore billed accordingly.<br>
                <br>
              </i></div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              So, apart from the cores requested, <b>should I also try
                to calculate the RAM requirements?</b></div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              For example, I see that Bruno's "leonardo-dcgp.ini" file
              specifies:</div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              <code>memory = 494000</code></div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              And the respective submitscript also has this line:</div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              <code>#SBATCH --mem 494000MB</code></div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              <br>
            </div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              I note that each node in Leonardo has 512GB of RAM, so
              that means that <b>the script requests ~94.2% of the RAM.</b></div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              I am not sure I follow the reasoning behind this.<br>
              <br>
            </div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              What is the default behavior of SimFactory if I were to
              remove the above specifications from the config files?</div>
            <div class="x_x_elementToProof"
style="font-family:Aptos,Aptos_EmbeddedFont,Aptos_MSFontService,Calibri,Helvetica,sans-serif; font-size:12pt; color:rgb(0,0,0)">
              Because, if by default Simfactory requests/uses all the
              RAM available in a node, then as far as I understand, it
              does not make sense to request fewer cores than a full
              node.<br>
              Let me know what you think.<br>
              <br>
              Best,<br>
              Panayotis <br>
              <br>
              <br>
            </div>
            <hr tabindex="-1" style="display:inline-block; width:98%">
            <div id="x_x_divRplyFwdMsg" dir="ltr"><font
                face="Calibri, sans-serif" color="#000000"
                style="font-size:11pt"><b>From:</b> Roland Haas
                <a class="x_moz-txt-link-rfc2396E"
                  href="mailto:rhaas@mail.ubc.ca" moz-do-not-send="true"><rhaas@mail.ubc.ca></a><br>
                <b>Sent:</b> Friday, September 26, 2025 4:31 PM<br>
                <b>To:</b> Bruno Giacomazzo <a
                  class="x_moz-txt-link-rfc2396E"
                  href="mailto:bruno.giacomazzo@unimib.it"
                  moz-do-not-send="true">
                  <bruno.giacomazzo@unimib.it></a><br>
                <b>Cc:</b> IOSIF PANAGIOTIS <a
                  class="x_moz-txt-link-rfc2396E"
                  href="mailto:PANAGIOTIS.IOSIF@units.it"
                  moz-do-not-send="true">
                  <PANAGIOTIS.IOSIF@units.it></a>; Einstein
                Toolkit Users <a class="x_moz-txt-link-rfc2396E"
                  href="mailto:users@einsteintoolkit.org"
                  moz-do-not-send="true">
                  <users@einsteintoolkit.org></a><br>
                <b>Subject:</b> Re: [Users] Inconsistency warnings:
                cores/threads mismatch [Leonardo cluster]</font>
              <div> </div>
            </div>
            <div class="x_x_BodyFragment"><font size="2"><span
                  style="font-size:11pt">
                  <div class="x_x_PlainText">Hello all,<br>
                    <br>
                    > I never used --cores and I don't know the
                    difference with procs.<br>
                    <br>
                    --cores is a synonym for --procs in simfactory. The
                    hope was to avoid<br>
                    the confusion of "procs" being "Processes" or
                    "Processors". Though it<br>
                    has been pointed out that the best name would
                    actually be "--threads"<br>
                    since that is what simfactory actually starts, which
                    then collides with<br>
                    "--num-threads" (threads per process).<br>
                    <br>
                    Does Leonardo actually charge you for partial nodes
                    if you do no use a<br>
                    full one? Simfactory is mostly written under the
                    assumption (true at<br>
                    the time) that HPC systems would give you full nodes
                    all the time, so<br>
                    if you use 1 core or 112 cores of a node, the charge
                    would be the same<br>
                    (though shared node systems are becoming more common
                    for HPC now [or<br>
                    again]).<br>
                    <br>
                    Yours,<br>
                    Roland<br>
                    <br>
                    -- <br>
                    My email is as private as my paper mail. I therefore
                    support encrypting<br>
                    and signing email messages. Get my PGP key from <a
                      href="http://pgp.mit.edu"
class="x_moz-txt-link-freetext moz-txt-link-freetext"
                      moz-do-not-send="true">
                      http://pgp.mit.edu</a> .<br>
                  </div>
                </span></font></div>
          </div>
          <br>
          <fieldset class="x_moz-mime-attachment-header"></fieldset>
          <pre class="x_moz-quote-pre">_______________________________________________
Users mailing list
<a class="x_moz-txt-link-abbreviated moz-txt-link-freetext"
          href="mailto:Users@einsteintoolkit.org" moz-do-not-send="true">Users@einsteintoolkit.org</a>
<a class="x_moz-txt-link-freetext moz-txt-link-freetext"
          href="http://lists.einsteintoolkit.org/mailman/listinfo/users"
          moz-do-not-send="true">http://lists.einsteintoolkit.org/mailman/listinfo/users</a>
</pre>
        </blockquote>
      </div>
      <br>
      <fieldset class="moz-mime-attachment-header"></fieldset>
      <pre wrap="" class="moz-quote-pre">_______________________________________________
Users mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Users@einsteintoolkit.org">Users@einsteintoolkit.org</a>
<a class="moz-txt-link-freetext" href="http://lists.einsteintoolkit.org/mailman/listinfo/users">http://lists.einsteintoolkit.org/mailman/listinfo/users</a>
</pre>
    </blockquote>
  </body>
</html>