<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <br>
    <div class="moz-cite-prefix">On 1/3/2018 12:10 AM, Matthew Knepley
      wrote:<br>
    </div>
    <blockquote type="cite"
cite="mid:CAMYG4Gnd-z9aL78nwyfGRSGSKs7saQMz3hubhzTmX_ND-hTZKQ@mail.gmail.com">
      <div dir="ltr">
        <div class="gmail_extra">
          <div class="gmail_quote">On Wed, Feb 28, 2018 at 10:45 AM, TAY
            wee-beng <span dir="ltr"><<a
                href="mailto:zonexo@gmail.com" target="_blank"
                moz-do-not-send="true">zonexo@gmail.com</a>></span>
            wrote:<br>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi,<br>
              <br>
              I have a CFD code which uses PETSc and HYPRE. I found that
              for a certain case with grid size of 192,570,048, I
              encounter scaling problem when my cores > 600. At 600
              cores, the code took 10min for 100 time steps. At 960,
              1440 and 2880 cores, it still takes around 10min. At 360
              cores, it took 15min.<br>
              <br>
              So how can I find the bottleneck? Any recommended steps?<br>
            </blockquote>
            <div><br>
            </div>
            <div>For any performance question, we need to see the output
              of -log_view for all test cases.</div>
          </div>
        </div>
      </div>
    </blockquote>
    Hi,<br>
    <br>
    To be more specific, I use PETSc KSPBCGS and HYPRE geometric
    multigrid (entirely based on HYPRE, no PETSc) for the momentum and
    Poisson eqns in my code.<br>
    <br>
    So can log_view be used in this case to give a meaningful? Since
    part of the code uses HYPRE?<br>
    <br>
    I also program another subroutine in the past which uses PETSc to
    solve the Poisson eqn. It uses either HYPRE's boomeramg, KSPBCGS or
    KSPGMRES.<br>
    <br>
    If I use boomeramg, can log_view be used in this case?<br>
    <br>
    Or do I have to use KSPBCGS or KSPGMRES, which is directly from
    PETSc? However, I ran KSPGMRES yesterday with the Poisson eqn and my
    ans didn't converge.<br>
    <br>
    Thanks.<br>
    <blockquote type="cite"
cite="mid:CAMYG4Gnd-z9aL78nwyfGRSGSKs7saQMz3hubhzTmX_ND-hTZKQ@mail.gmail.com">
      <div dir="ltr">
        <div class="gmail_extra">
          <div class="gmail_quote">
            <div> </div>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">
              I must also mention that I partition my grid only in the x
              and y direction. There is no partitioning in the z
              direction due to limited code development. I wonder if
              there is a strong effect in this case.</blockquote>
            <div><br>
            </div>
            <div>Maybe. Usually what happens is you fill up memory with
              a z-column and cannot scale further.</div>
            <div><br>
            </div>
            <div>  Thanks,</div>
            <div><br>
            </div>
            <div>     Matt</div>
            <div> </div>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex"><span
                class="HOEnZb"><font color="#888888"><br>
                  -- <br>
                  Thank you very much<br>
                  <br>
                  Yours sincerely,<br>
                  <br>
                  ==============================<wbr>==================<br>
                  TAY Wee-Beng 郑伟明 (Zheng Weiming)<br>
                  Personal research webpage: <a
                    href="http://tayweebeng.wixsite.com/website"
                    rel="noreferrer" target="_blank"
                    moz-do-not-send="true">http://tayweebeng.wixsite.com/<wbr>website</a><br>
                  Youtube research showcase: <a
                    href="https://www.youtube.com/channel/UC72ZHtvQNMpNs2uRTSToiLA"
                    rel="noreferrer" target="_blank"
                    moz-do-not-send="true">https://www.youtube.com/channe<wbr>l/UC72ZHtvQNMpNs2uRTSToiLA</a><br>
                  linkedin: <a
                    href="http://www.linkedin.com/in/tay-weebeng"
                    rel="noreferrer" target="_blank"
                    moz-do-not-send="true">www.linkedin.com/in/tay-weeben<wbr>g</a><br>
                  ==============================<wbr>==================<br>
                  <br>
                </font></span></blockquote>
          </div>
          <br>
          <br clear="all">
          <div><br>
          </div>
          -- <br>
          <div class="gmail_signature" data-smartmail="gmail_signature">
            <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.caam.rice.edu/%7Emk51/"
                      target="_blank" moz-do-not-send="true">https://www.cse.buffalo.edu/~knepley/</a><br>
                  </div>
                </div>
              </div>
            </div>
          </div>
        </div>
      </div>
    </blockquote>
    <br>
  </body>
</html>