<div dir="ltr"><br><div class="gmail_extra"><br><div class="gmail_quote">On Thu, May 4, 2017 at 2:27 PM, Jiri Gaisler <span dir="ltr"><<a href="mailto:jiri@gaisler.se" target="_blank">jiri@gaisler.se</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div bgcolor="#FFFFFF" text="#000000"><div><div class="h5">
<p><br>
</p>
<br>
<div class="m_7096908910262491930moz-cite-prefix">On 05/04/2017 04:00 PM, Joel Sherrill
wrote:<br>
</div>
<blockquote 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 href="mailto:sebastian.huber@embedded-brains.de" target="_blank">sebastian.huber@embedded-<wbr>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="m_7096908910262491930gmail-">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="m_7096908910262491930gmail-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></div></div>
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 ...<span class="HOEnZb"><font color="#888888"><br>
<br></font></span></div></blockquote><div>Thanks. I wonder if we should have a leon3 variant with hardware FPU</div><div>enabled.</div><div><br></div><div>--joel</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div bgcolor="#FFFFFF" text="#000000"><span class="HOEnZb"><font color="#888888">
Jiri.<br>
<br>
<br>
</font></span></div>
</blockquote></div><br></div></div>