Re: Substituting Checksum Algorithm (was: Enabling Checksums) - Mailing list pgsql-hackers

From Jeff Davis
Subject Re: Substituting Checksum Algorithm (was: Enabling Checksums)
Date
Msg-id 1367013190.11576.249.camel@sussancws0025
Whole thread Raw
In response to Re: Substituting Checksum Algorithm (was: Enabling Checksums)  (Greg Smith <greg@2ndQuadrant.com>)
List pgsql-hackers
On Fri, 2013-04-26 at 16:40 -0400, Greg Smith wrote:
> I think I need to do two baselines:  master without checksums, and 
> master with extra optimizations but still without checksums.  It may be 
> the case that using better compile time optimizations gives a general 
> speedup that's worth considering regardless.  The optimizations seem to 
> have a very significant impact on the checksum feature, but I'd like to 
> quantify how they change the code a little bit before even getting into 
> that.

The patch only affects optimization flags used when compiling
checksum.c, so it should have no effect on other areas of the code.

If you want to compile the whole source with those flags, then just do:
 CFLAGS="-msse4.1 -funroll-loops -ftree-vectorize" ./configure

Changing the optimization flags for existing code will have a larger
impact and should be considered separately from checksums.

Regards,Jeff Davis





pgsql-hackers by date:

Previous
From: Greg Smith
Date:
Subject: Re: Substituting Checksum Algorithm (was: Enabling Checksums)
Next
From: Jeff Janes
Date:
Subject: pg_ctl non-idempotent behavior change