<p dir="ltr"><br /> Em 12/09/2014 17:23, "Andres Freund" <<a
href="mailto:andres@2ndquadrant.com">andres@2ndquadrant.com</a>>escreveu:<br /> ><br /> > On 2014-09-12
23:03:00+0300, Heikki Linnakangas wrote:<br /> > > On 09/12/2014 10:54 PM, Abhijit Menon-Sen wrote:<br /> >
>>At 2014-09-12 22:38:01 +0300, <a href="mailto:hlinnakangas@vmware.com">hlinnakangas@vmware.com</a> wrote:<br />
>> >><br /> > > >>We probably should consider switching to a faster CRC algorithm again,<br />
>> >>regardless of what we do with compression.<br /> > > ><br /> > > >As it happens, I'm
alreadyworking on resurrecting a patch that Andres<br /> > > >posted in 2010 to switch to zlib's faster CRC
implementation.<br/> > ><br /> > > As it happens, I also wrote an implementation of Slice-by-4 the other
day<br/> > > :-). Haven't gotten around to post it, but here it is.<br /> > ><br /> > > What
algorithmdoes zlib use for CRC calculation?<br /> ><br /> > Also slice-by-4, with a manually unrolled loop doing
32bytesat once, using<br /> > individual slice-by-4's. IIRC I tried and removing that slowed things<br /> > down
overall.What it also did was move crc to a function. I'm not sure<br /> > why I did it that way, but it really might
bebeneficial - if you look<br /> > at profiles today there's sometimes icache/decoding stalls...<br /> ><br />
>Hm. Let me look:<br /> > <a
href="http://archives.postgresql.org/message-id/201005202227.49990.andres%40anarazel.de">http://archives.postgresql.org/message-id/201005202227.49990.andres%40anarazel.de</a><br
/>><br /> > Ick, there's quite some debugging leftovers ;)<br /> ><br /> > I think it might be a good idea
toalso switch the polynom at the same<br /> > time. I really really think we should, when the hardware supports,
use<br/> > the polynom that's available in SSE4.2. It has similar properties, can<br /> > implemented in software
justthe same...<br /> ><br /> > Greetings,<br /> ><br /> > Andres Freund<br /> ><br /> > --<br />
> Andres Freund <a href="http://www.2ndQuadrant.com/">http://www.2ndQuadrant.com/</a><br /> >
PostgreSQLDevelopment, 24x7 Support, Training & Services<br /> ><br /> ><br /> > --<br /> > Sent via
pgsql-hackersmailing list (<a href="mailto:pgsql-hackers@postgresql.org">pgsql-hackers@postgresql.org</a>)<br /> >
Tomake changes to your subscription:<br /> > <a
href="http://www.postgresql.org/mailpref/pgsql-hackers">http://www.postgresql.org/mailpref/pgsql-hackers</a><p
dir="ltr">ThisGoogle library is worth a look <a
href="https://code.google.com/p/crcutil/">https://code.google.com/p/crcutil/</a>as it has some extremely optimized
versions.