<div dir="ltr"><div><div><div>Sudo was in place.<br><br>But somehow it still had multiple cores in the shield.<br></div>Now it must be just 1 core. Should make the difference with variations but slower to test.<br><br></div>Thanks<br><br></div>Galina<br><div><div><br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Mon, Mar 26, 2018 at 1:16 PM, Galina Kistanova <span dir="ltr"><<a href="mailto:gkistanova@gmail.com" target="_blank">gkistanova@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div><div><div><div>Hi Rafael,<br><br></div>Thanks for mentioning this.<br><br></div>It should be running as root, but I'll double check anyway.<br><br></div>Thanks<span class="HOEnZb"><font color="#888888"><br><br></font></span></div><span class="HOEnZb"><font color="#888888">Galina<br></font></span></div><div class="HOEnZb"><div class="h5"><div class="gmail_extra"><br><div class="gmail_quote">On Mon, Mar 26, 2018 at 10:47 AM, Rafael Avila de Espindola <span dir="ltr"><<a href="mailto:rafael.espindola@gmail.com" target="_blank">rafael.espindola@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span>Galina Kistanova <<a href="mailto:gkistanova@gmail.com" target="_blank">gkistanova@gmail.com</a>> writes:<br>
<br>
> Disabling swap and having a single CPU in the shield group didn't change<br>
> much, besides cpu-migrations and context-switches, which now are 0<br>
> obviously.<br>
> That clustering remains the same. It is also stable to the number of runs<br>
> (I have changed the test to run 20 times in the middle of that range on the<br>
> right).<br>
><br>
> <a href="http://lnt.llvm.org/db_default/v4/link/graph?highlight_run=426&plot.9=1.9.6" rel="noreferrer" target="_blank">http://lnt.llvm.org/db_default<wbr>/v4/link/graph?highlight_run=<wbr>426&plot.9=1.9.6</a><br>
><br>
> Unless somebody has a good idea what else we should try, it seems we have<br>
> it as good as it could be with the current approach.<br>
<br>
</span>I am benchmarking a patch and just remembered one thing: cset will only<br>
enable the shield if run as root, but it will still run the program<br>
outside otherwise.<br>
<br>
Is the benchmark being run as root? if not that might explain the large<br>
variations from run to run.<br>
<br>
Cheers,<br>
Rafael<br>
</blockquote></div><br></div>
</div></div></blockquote></div><br></div></div>