<html>
  <head>
    <meta content="text/html; charset=windows-1252"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <p><br>
    </p>
    <br>
    <div class="moz-cite-prefix">On 05/04/2017 04:00 PM, Joel Sherrill
      wrote:<br>
    </div>
    <blockquote
cite="mid:CAF9ehCVUxCiEzvAqus3db1sH3q_2S4+F33MmvNr7y+Uz86EF2A@mail.gmail.com"
      type="cite">
      <div dir="ltr"><br>
        <div class="gmail_extra"><br>
          <div class="gmail_quote">On Thu, May 4, 2017 at 12:34 AM,
            Sebastian Huber <span dir="ltr"><<a
                moz-do-not-send="true"
                href="mailto:sebastian.huber@embedded-brains.de"
                target="_blank">sebastian.huber@embedded-brains.de</a>></span>
            wrote:<br>
            <blockquote class="gmail_quote" style="margin:0px 0px 0px
              0.8ex;border-left:1px solid
              rgb(204,204,204);padding-left:1ex"><span class="gmail-">On
                03/05/17 22:52, Joel Sherrill wrote:<br>
                <blockquote class="gmail_quote" style="margin:0px 0px
                  0px 0.8ex;border-left:1px solid
                  rgb(204,204,204);padding-left:1ex">
                  Has anyone run the whetstone benchmark on real leon3
                  HW?<br>
                  I'm almost at 20 minutes of simulated CPU time on tsim
                  so<br>
                  am suspicious something is wrong.<br>
                  <br>
                  Any ideas?<br>
                </blockquote>
                <br>
              </span>
              Maybe soft-float plus a low CPU frequency? Do you have a
              function profiling histogram of the execution?<span
                class="gmail-HOEnZb"><font color="#888888"><br>
                  <br>
                </font></span></blockquote>
            <div><br>
            </div>
            <div>It is the default leon3 build and the leon3 is
              simulated at 50Mhz.  I have </div>
            <div>never used profiling on tsim. Any guidance?</div>
            <div><br>
            </div>
            <div>The test  did finally finish and here is the output of
              tsim's perf command:</div>
            <div><br>
            </div>
            <div>
              <div>*** END OF TEST WHETSTONE ***</div>
              <div><br>
              </div>
              <div>Program exited normally.</div>
              <div>tsim> perf</div>
              <div>                                                     
                                          </div>
              <div> Cycles       : 74462444473</div>
              <div> Instructions : 45632185502</div>
              <div> Overall CPI  :      1.63</div>
              <div><br>
              </div>
              <div> CPU performance (50.0 MHz)  : 30.64 MOPS (30.64
                MIPS,  0.00 MFLOPS)</div>
              <div> Cache hit rate              : 98.0 % (inst: 97.9,
                data:99.7)</div>
              <div> AHB bandwidth utilisation   : 34.8 % (inst: 14.1,
                data: 20.8)</div>
              <div> Simulated time              : 1489.25 s</div>
              <div> Processor utilisation       : 100.00 %</div>
              <div> Real-time performance       : 122.02 %</div>
              <div> Simulator performance       :  37.39 MIPS</div>
              <div> Used time (sys + user)      : 1220.46 s</div>
            </div>
            <div><br>
            </div>
            <div>So it just took a LONG time. </div>
            <div><br>
            </div>
            <div>--joel</div>
          </div>
        </div>
      </div>
    </blockquote>
    <br>
    Note that TSIM reports 0.0 MFLOPS. This means that soft-float was
    used to compile the benchmark. Using hard-float should give about
    100x speed-up ...<br>
    <br>
    Jiri.<br>
    <br>
    <br>
  </body>
</html>