Re: Move PinBuffer and UnpinBuffer to atomics

<div dir="ltr"><div class="gmail_extra"><br /><div class="gmail_quote">On Fri, Mar 25, 2016 at 8:09 PM, Alexander
Korotkov<span dir="ltr"><<a href="mailto:a.korotkov@postgrespro.ru"
target="_blank">a.korotkov@postgrespro.ru</a>></span>wrote:<br /><blockquote class="gmail_quote" style="margin:0px
0px0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">Could
anybodyrun benchmarks?  Feature freeze is soon, but it would be *very nice* to fit it into 9.6 release cycle, because
itgreatly improves scalability on large machines.  Without this patch PostgreSQL 9.6 will be significantly behind
competitorslike MySQL 5.7.</blockquote></div><br />I have run the performance and here are the results.. With latest
patchI did not see any regression at lower client count (median of 3 reading).</div><div class="gmail_extra"><br
/></div><divclass="gmail_extra">scale factor 1000 shared buffer 8GB readonly<span class="" style="white-space:pre">
</span><br/></div><div class="gmail_extra"><div class="gmail_extra"><span class="" style="white-space:pre">
</span></div><divclass="gmail_extra"><b>Client<span class="" style="white-space:pre"> </span>Base<span class=""
style="white-space:pre"></span>patch</b></div><div class="gmail_extra">1<span class="" style="white-space:pre">
</span>12957<spanclass="" style="white-space:pre"> </span>13068</div><div class="gmail_extra">2<span class=""
style="white-space:pre"></span>24931<span class="" style="white-space:pre"> </span>25816</div><div
class="gmail_extra">4<spanclass="" style="white-space:pre"> </span>46311<span class="" style="white-space:pre">
</span>48767</div><divclass="gmail_extra">32<span class="" style="white-space:pre"> </span>300921<span class=""
style="white-space:pre"></span>310062</div><div class="gmail_extra">64<span class="" style="white-space:pre">
</span>387623<spanclass="" style="white-space:pre"> </span>493843</div><div class="gmail_extra">128<span class=""
style="white-space:pre"></span>249635<span class="" style="white-space:pre"> </span>583513</div><div
class="gmail_extra"><spanclass="" style="white-space:pre"> </span></div><div class="gmail_extra"><span class=""
style="white-space:pre"></span></div><div class="gmail_extra">scale factor 300 shared buffer 8GB readonly<span class=""
style="white-space:pre"></span></div><div class="gmail_extra"><span class="" style="white-space:pre"> </span></div><div
class="gmail_extra"><b>Client<spanclass="" style="white-space:pre"> </span>Base<span class="" style="white-space:pre">
</span>patch</b></div><divclass="gmail_extra">1<span class="" style="white-space:pre"> </span>14537<span class=""
style="white-space:pre"></span>14586    --> one thread number looks little less, generally I get ~18000 (will
recheck).</div><divclass="gmail_extra">2<span class="" style="white-space:pre"> </span>34703<span class=""
style="white-space:pre"></span>33929    --> may be run to run variance (once I get time, will recheck)</div><div
class="gmail_extra">4<spanclass="" style="white-space:pre"> </span>67744<span class="" style="white-space:pre">
</span>69069</div><divclass="gmail_extra">32<span class="" style="white-space:pre"> </span>312575<span class=""
style="white-space:pre"></span>336012</div><div class="gmail_extra">64<span class="" style="white-space:pre">
</span>213312<spanclass="" style="white-space:pre"> </span>539056</div><div class="gmail_extra">128<span class=""
style="white-space:pre"></span>190139<span class="" style="white-space:pre"> </span>380122</div><div
class="gmail_extra"><br/></div><div class="gmail_extra"><b>Summary:</b></div><div class="gmail_extra"><br /></div><div
class="gmail_extra">Actuallywith 64 client we have seen ~470,000 TPS with head also, by revering commit
6150a1b0.</div><divclass="gmail_extra">refer this thread: (<a
href="http://www.postgresql.org/message-id/CAA4eK1+ZeB8PMwwktf+3bRS0Pt4Ux6Rs6Aom0uip8c6shJWmyg@mail.gmail.com">http://www.postgresql.org/message-id/CAA4eK1+ZeB8PMwwktf+3bRS0Pt4Ux6Rs6Aom0uip8c6shJWmyg@mail.gmail.com</a>)</div><div
class="gmail_extra"><br/></div><div class="gmail_extra">I haven't tested this patch by reverting commit 6150a1b0, so
notsure can this patch give even better performance ?</div><div class="gmail_extra"><br /></div><div
class="gmail_extra">Italso points to the case, what Andres has mentioned in this thread.</div><div
class="gmail_extra"><br/></div><div class="gmail_extra"><a
href="http://www.postgresql.org/message-id/20160226191158.3vidtk3ktcmhimdu@alap3.anarazel.de">http://www.postgresql.org/message-id/20160226191158.3vidtk3ktcmhimdu@alap3.anarazel.de</a></div><div
class="gmail_extra"><spanstyle="color:rgb(80,0,80);font-size:12.8px"><br /></span></div><div class="gmail_extra"><span
style="color:rgb(80,0,80);font-size:12.8px">Regards,</span><br/></div></div><div class="gmail_extra"><div
class="gmail_signature"><divdir="ltr"><span style="color:rgb(80,0,80);font-size:12.8px">Dilip Kumar</span><br
style="color:rgb(80,0,80);font-size:12.8px"/><span style="color:rgb(80,0,80);font-size:12.8px">EnterpriseDB: </span><a
href="http://www.enterprisedb.com/"style="color:rgb(17,85,204);font-size:12.8px"
target="_blank">http://www.enterprisedb.com</a><br/></div></div></div></div>