On Wed, Aug 12, 2015 at 04:47:55PM -0400, Robert Haas wrote:
> On Wed, Aug 12, 2015 at 4:34 PM, Heikki Linnakangas <hlinnaka@iki.fi> wrote:
> > Andres didn't mention how big the performance benefit he saw with pgbench
> > was, but I bet it was barely distinguishible from noise. But that's OK. In
> > fact, there's no reason to believe this would make any difference to
> > performance. The point is to make the code more readable, and it certainly
> > achieves that.
>
> I think that when Bruce macro-ized this ten years ago or whenever it
> was, he got a significant performance benefit from it; otherwise I
> don't think he would have done it.
(You over-estimate me. ;-) )
What happened is that I was looking at call graph counts and
fastgetattr() was called a bazillion times, so I inlined it, and saw a
noticeably performance improvement, maybe 2% on an in-memory
SELECT-only workload. Same with a few other macros I created in those
early years.
Frankly, my hacks last a lot longer than I expected. (Did someone say
pg_upgrade. :-) )
-- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB
http://enterprisedb.com
+ Everyone has their own god. +