<html>
  <head>
    <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    Tony,<br>
    <br>
    Are you setup to run specjbb2013?  I did a run with<br>
    <br>
    the PerStride changes (original patch) vs a baseline and<br>
    saw an increase in ParNew times.  I attached a plot of the <br>
    pause times.  I used<br>
    <br>
    -server -Xmx30g -Xms30g -Xmn20g -XX:+UseConcMarkSweepGC
    -XX:+UseLargePages<br>
    <br>
    This was run on a linux system with 8 cores.<br>
    <br>
    From a run of specjbb2005 on the same system there is an improvement<br>
    in the ParNew times.<br>
    <br>
    I'll repeat the experiment with the latest patch.   If you can try
    to repeat the experiment,<br>
    that would give me some confidence that it's not just me.<br>
    <br>
    Thanks.<br>
    <br>
    Jon<br>
    <br>
    <br>
    <div class="moz-cite-prefix">On 07/02/2015 12:39 PM, Tony Printezis
      wrote:<br>
    </div>
    <blockquote cite="mid:etPan.5595937a.283b46fd.386@tw-mbp-tprintezis"
      type="cite">
      <style>body{font-family:Helvetica,Arial;font-size:13px}</style>
      <div id="bloop_customfont"
        style="font-family:Helvetica,Arial;font-size:13px; color:
        rgba(0,0,0,1.0); margin: 0px; line-height: auto;">Thomas,</div>
      <div id="bloop_customfont"
        style="font-family:Helvetica,Arial;font-size:13px; color:
        rgba(0,0,0,1.0); margin: 0px; line-height: auto;"><br>
      </div>
      <div id="bloop_customfont"
        style="font-family:Helvetica,Arial;font-size:13px; color:
        rgba(0,0,0,1.0); margin: 0px; line-height: auto;">Latest
        version:</div>
      <div id="bloop_customfont"
        style="font-family:Helvetica,Arial;font-size:13px; color:
        rgba(0,0,0,1.0); margin: 0px; line-height: auto;"><br>
      </div>
      <div id="bloop_customfont"
        style="font-family:Helvetica,Arial;font-size:13px; color:
        rgba(0,0,0,1.0); margin: 0px; line-height: auto;"><a
          moz-do-not-send="true"
          href="http://cr.openjdk.java.net/%7Etonyp/8086056/webrev.3/">http://cr.openjdk.java.net/~tonyp/8086056/webrev.3/</a></div>
      <div id="bloop_customfont"
        style="font-family:Helvetica,Arial;font-size:13px; color:
        rgba(0,0,0,1.0); margin: 0px; line-height: auto;"><br>
      </div>
      <div id="bloop_customfont"
        style="font-family:Helvetica,Arial;font-size:13px; color:
        rgba(0,0,0,1.0); margin: 0px; line-height: auto;">See below for
        comments.</div>
      <br>
      <p class="airmail_on" style="color:#000;">On July 2, 2015 at
        4:43:42 AM, Thomas Schatzl (<a moz-do-not-send="true"
          href="mailto:thomas.schatzl@oracle.com">thomas.schatzl@oracle.com</a>)
        wrote:</p>
      <div>
        <blockquote type="cite" class="clean_bq" style="color: rgb(0, 0,
          0); font-family: Helvetica, Arial; font-size: 13px;
          font-style: normal; font-variant: normal; font-weight: normal;
          letter-spacing: normal; line-height: normal; orphans: auto;
          text-align: start; text-indent: 0px; text-transform: none;
          white-space: normal; widows: auto; word-spacing: 0px;
          -webkit-text-stroke-width: 0px;"><span>
            <div>
              <div>Hi Tony,<span class="Apple-converted-space"> </span><br>
                <br>
                On Wed, 2015-07-01 at 18:05 -0400, Tony Printezis wrote:<span
                  class="Apple-converted-space"> </span><br>
                > Latest changes, as discussed with Bengt and Thomas,<span
                  class="Apple-converted-space"> </span><br>
                ><span class="Apple-converted-space"> </span><br>
                > <a class="moz-txt-link-freetext" href="http://cr.openjdk.java.net/~tonyp/8086056/webrev.2/">http://cr.openjdk.java.net/~tonyp/8086056/webrev.2/</a><span
                  class="Apple-converted-space"> </span><br>
                <br>
                a few suggestions:<span class="Apple-converted-space"> </span><br>
                <br>
                - please use braces even for simple statements in
                if-clauses, e.g.<span class="Apple-converted-space"> </span></div>
            </div>
          </span></blockquote>
      </div>
      <div>
        <div>
          <blockquote type="cite" class="clean_bq" style="color: rgb(0,
            0, 0); font-family: Helvetica, Arial; font-size: 13px;
            font-style: normal; font-variant: normal; font-weight:
            normal; letter-spacing: normal; line-height: normal;
            orphans: auto; text-align: start; text-indent: 0px;
            text-transform: none; white-space: normal; widows: auto;
            word-spacing: 0px; -webkit-text-stroke-width: 0px;"><span>
              <div>
                <div><br>
                  parNewGeneration.cpp:<span
                    class="Apple-converted-space"> </span><br>
                  <br>
                  883 if (!UseDynamicParGCStrides) return;<span
                    class="Apple-converted-space"> </span></div>
              </div>
            </span></blockquote>
        </div>
        <p><br>
        </p>
        <p>I did that. But I actually added an extra verbose statement
          in the body of the if-statement (I thought it’d be helpful to
          also print the value of ParGCCardsPerStrideChunk when the
          dynamic calculation is off). So, I don’t have a simple return
          any more, but I’ll keep that in mind in the future (was it
          anywhere else? I didn’t immediately spot another occurrence of
          that).</p>
        <p><br>
        </p>
        <div>
          <div>
            <blockquote type="cite" class="clean_bq" style="color:
              rgb(0, 0, 0); font-family: Helvetica, Arial; font-size:
              13px; font-style: normal; font-variant: normal;
              font-weight: normal; letter-spacing: normal; line-height:
              normal; orphans: auto; text-align: start; text-indent:
              0px; text-transform: none; white-space: normal; widows:
              auto; word-spacing: 0px; -webkit-text-stroke-width: 0px;"><span>
                <div>
                  <div><br>
                    - Comments should start with upper case:<span
                      class="Apple-converted-space"> </span><br>
                    <br>
                    parNewGeneration.hpp:<span
                      class="Apple-converted-space"> </span><br>
                    <br>
                    354 // automatically calculate
                    ParGCCardsPerStrideChunk based on the<span
                      class="Apple-converted-space"> </span><br>
                    old<span class="Apple-converted-space"> </span></div>
                </div>
              </span></blockquote>
          </div>
          <p><br>
          </p>
          <p>Done.</p>
          <p><br>
          </p>
          <div>
            <div>
              <blockquote type="cite" class="clean_bq" style="color:
                rgb(0, 0, 0); font-family: Helvetica, Arial; font-size:
                13px; font-style: normal; font-variant: normal;
                font-weight: normal; letter-spacing: normal;
                line-height: normal; orphans: auto; text-align: start;
                text-indent: 0px; text-transform: none; white-space:
                normal; widows: auto; word-spacing: 0px;
                -webkit-text-stroke-width: 0px;"><span>
                  <div>
                    <div><br>
                      - parameter calculation in
                      ParNewGeneration::adjust_cards_per_stride:<span
                        class="Apple-converted-space"> </span><br>
                      - DynamicParGCStridesMinSize should be > 0
                      (only if !<span class="Apple-converted-space"> </span><br>
                      UseDynamicParGCStrides is enabled, not sure if
                      that can be encoded<span
                        class="Apple-converted-space"> </span><br>
                      within the framework)<span
                        class="Apple-converted-space"> </span></div>
                  </div>
                </span></blockquote>
            </div>
            <p><br>
            </p>
            <p>If !UseDynamicParGCStrides, then both sets of min / max
              parameters are not used. I think it’d make the constraint
              checking a pain having to sanity-check parameters whether
              a flag is on or off (unless they mean different things in
              either case, of course).</p>
            <p>I added a range check to the min / max stride size args,
              added a lower bound of 1, and decided to add an artificial
              upper bound of 32K (I have tested with up to 8K). Is that
              reasonable? We can revisit that if we need to increse it.
              Note that you can still set ParGCCardsPerStrideChunk to
              something higher if required.</p>
            <p><br>
            </p>
            <div>
              <div>
                <blockquote type="cite" class="clean_bq" style="color:
                  rgb(0, 0, 0); font-family: Helvetica, Arial;
                  font-size: 13px; font-style: normal; font-variant:
                  normal; font-weight: normal; letter-spacing: normal;
                  line-height: normal; orphans: auto; text-align: start;
                  text-indent: 0px; text-transform: none; white-space:
                  normal; widows: auto; word-spacing: 0px;
                  -webkit-text-stroke-width: 0px;"><span>
                    <div>
                      <div><br>
                        - the issues with the Dynamic*Capacity variables
                        Sangheon mentioned,<span
                          class="Apple-converted-space"> </span><br>
                        although I think it is fine that they are equal,
                        just that "r" needs to<span
                          class="Apple-converted-space"> </span><br>
                        be set correctly.<span
                          class="Apple-converted-space"> </span></div>
                    </div>
                  </span></blockquote>
              </div>
              <p><br>
              </p>
              <p>Yeah, fixed. (See previous e-mail.)</p>
              <p><br>
              </p>
              <div>
                <div>
                  <blockquote type="cite" class="clean_bq" style="color:
                    rgb(0, 0, 0); font-family: Helvetica, Arial;
                    font-size: 13px; font-style: normal; font-variant:
                    normal; font-weight: normal; letter-spacing: normal;
                    line-height: normal; orphans: auto; text-align:
                    start; text-indent: 0px; text-transform: none;
                    white-space: normal; widows: auto; word-spacing:
                    0px; -webkit-text-stroke-width: 0px;"><span>
                      <div>
                        <div><br>
                          Just wanting to check back if it was intended
                          that the code always<span
                            class="Apple-converted-space"> </span><br>
                          rounds down, achieving the maximum stride only
                          at the end of the<span
                            class="Apple-converted-space"> </span><br>
                          interval. E.g.<span
                            class="Apple-converted-space"> </span><br>
                          <br>
                          912 const int stride_size_log =
                          log2_long((jlong) stride_size);<span
                            class="Apple-converted-space"> </span><br>
                          913 res = (size_t) 1 << stride_size_log;<span
                            class="Apple-converted-space"> </span><br>
                          <br>
                          It's fine with me, just asking.<span
                            class="Apple-converted-space"> </span></div>
                      </div>
                    </span></blockquote>
                </div>
                <p><br>
                </p>
                <p>I did the simplest calculation. I can revise it if
                  you want. Also note that if capacity >
                  max_capacity, then stride size will be set to the max
                  stride size.</p>
                <p>Tony</p>
                <p><br>
                </p>
                <div>
                  <blockquote type="cite" class="clean_bq" style="color:
                    rgb(0, 0, 0); font-family: Helvetica, Arial;
                    font-size: 13px; font-style: normal; font-variant:
                    normal; font-weight: normal; letter-spacing: normal;
                    line-height: normal; orphans: auto; text-align:
                    start; text-indent: 0px; text-transform: none;
                    white-space: normal; widows: auto; word-spacing:
                    0px; -webkit-text-stroke-width: 0px;"><span>
                      <div>
                        <div><br>
                          Thanks,<span class="Apple-converted-space"> </span><br>
                          Thomas<span class="Apple-converted-space"> </span><br>
                          <br>
                          <br>
                          <br>
                        </div>
                      </div>
                    </span></blockquote>
                  <br class="Apple-interchange-newline">
                </div>
                <br class="Apple-interchange-newline">
              </div>
              <br class="Apple-interchange-newline">
            </div>
            <br class="Apple-interchange-newline">
          </div>
          <br class="Apple-interchange-newline">
        </div>
        <br class="Apple-interchange-newline">
      </div>
      <div id="bloop_sign_1435865248348184064" class="bloop_sign">
        <div style="font-family:helvetica,arial;font-size:13px">
          <div>-----</div>
          <div><br>
          </div>
          <div>Tony Printezis | JVM/GC Engineer / VM Team | Twitter</div>
          <div><br>
          </div>
          <div>@TonyPrintezis</div>
          <div><a moz-do-not-send="true"
              href="mailto:tprintezis@twitter.com">tprintezis@twitter.com</a></div>
          <div><br>
          </div>
        </div>
      </div>
    </blockquote>
    <br>
  </body>
</html>