<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <p>Hi Matt,</p>
    <p>I plotted the memory scalings using different threshold values.
      The two scalings are slightly translated (from -22 to -88 mB) but
      this gain is neglectable. The 3.6-scaling keeps being robust while
      the 3.10-scaling deteriorates.</p>
    <p>Do you have any other suggestion?</p>
    <p>Thanks<br>
    </p>
    Myriam <br>
    <br>
    <div class="moz-cite-prefix">Le 03/02/19 à 02:27, Matthew Knepley a
      écrit :<br>
    </div>
    <blockquote type="cite"
cite="mid:CAMYG4GkPNv4k501_+CpB2yXtxV812PgJLxPLKkvM3zx=3_mLjg@mail.gmail.com">
      <meta http-equiv="content-type" content="text/html; charset=utf-8">
      <div dir="ltr">
        <div dir="ltr">
          <div dir="ltr">On Fri, Mar 1, 2019 at 10:53 AM Myriam
            Peyrounette via petsc-users <<a
              href="mailto:petsc-users@mcs.anl.gov"
              moz-do-not-send="true">petsc-users@mcs.anl.gov</a>>
            wrote:<br>
          </div>
          <div class="gmail_quote">
            <blockquote class="gmail_quote" style="margin:0px 0px 0px
              0.8ex;border-left:1px solid
              rgb(204,204,204);padding-left:1ex">Hi,<br>
              <br>
              I used to run my code with PETSc 3.6. Since I upgraded the
              PETSc version<br>
              to 3.10, this code has a bad memory scaling.<br>
              <br>
              To report this issue, I took the PETSc script ex42.c and
              slightly<br>
              modified it so that the KSP and PC configurations are the
              same as in my<br>
              code. In particular, I use a "personnalised" multi-grid
              method. The<br>
              modifications are indicated by the keyword "TopBridge" in
              the attached<br>
              scripts.<br>
              <br>
              To plot the memory (weak) scaling, I ran four calculations
              for each<br>
              script with increasing problem sizes and computations
              cores:<br>
              <br>
              1. 100,000 elts on 4 cores<br>
              2. 1 million elts on 40 cores<br>
              3. 10 millions elts on 400 cores<br>
              4. 100 millions elts on 4,000 cores<br>
              <br>
              The resulting graph is also attached. The scaling using
              PETSc 3.10<br>
              clearly deteriorates for large cases, while the one using
              PETSc 3.6 is<br>
              robust.<br>
              <br>
              After a few tests, I found that the scaling is mostly
              sensitive to the<br>
              use of the AMG method for the coarse grid (line 1780 in<br>
              main_ex42_petsc36.cc). In particular, the performance
              strongly<br>
              deteriorates when commenting lines 1777 to 1790 (in
              main_ex42_petsc36.cc).<br>
              <br>
              Do you have any idea of what changed between version 3.6
              and version<br>
              3.10 that may imply such degradation?<br>
            </blockquote>
            <div><br>
            </div>
            <div>I believe the default values for PCGAMG changed between
              versions. It sounds like the coarsening rate</div>
            <div>is not great enough, so that these grids are too large.
              This can be set using:</div>
            <div><br>
            </div>
            <div>  <a
href="https://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/PC/PCGAMGSetThreshold.html"
                moz-do-not-send="true">https://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/PC/PCGAMGSetThreshold.html</a></div>
            <div><br>
            </div>
            <div>There is some explanation of this effect on that page.
              Let us know if setting this does not correct the
              situation.</div>
            <div><br>
            </div>
            <div>  Thanks,</div>
            <div><br>
            </div>
            <div>     Matt</div>
            <div> </div>
            <blockquote class="gmail_quote" style="margin:0px 0px 0px
              0.8ex;border-left:1px solid
              rgb(204,204,204);padding-left:1ex">
              Let me know if you need further information.<br>
              <br>
              Best,<br>
              <br>
              Myriam Peyrounette<br>
              <br>
              <br>
              -- <br>
              Myriam Peyrounette<br>
              CNRS/IDRIS - HLST<br>
              --<br>
              <br>
            </blockquote>
          </div>
          <br clear="all">
          <div><br>
          </div>
          -- <br>
          <div dir="ltr" class="gmail_signature">
            <div dir="ltr">
              <div>
                <div dir="ltr">
                  <div>
                    <div dir="ltr">
                      <div>What most experimenters take for granted
                        before they begin their experiments is
                        infinitely more interesting than any results to
                        which their experiments lead.<br>
                        -- Norbert Wiener</div>
                      <div><br>
                      </div>
                      <div><a
                          href="http://www.cse.buffalo.edu/%7Eknepley/"
                          target="_blank" moz-do-not-send="true">https://www.cse.buffalo.edu/~knepley/</a><br>
                      </div>
                    </div>
                  </div>
                </div>
              </div>
            </div>
          </div>
        </div>
      </div>
    </blockquote>
    <br>
    <pre class="moz-signature" cols="72">-- 
Myriam Peyrounette
CNRS/IDRIS - HLST
--
</pre>
  </body>
</html>