Re: performance regression when filling in a table

Hi,

On 2019-04-30 07:12:03 +0200, Fabien COELHO wrote:
> On my SSD Ubuntu laptop, with postgres-distributed binaries and unmodified
> default settings using local connections:

> ## pg 11.2
> > time pgbench -i -s 100
> ...
> done in 31.51 s
> # (drop tables 0.00 s, create tables 0.01 s, generate 21.30 s, vacuum 3.32 s, primary keys 6.88 s).
> # real 0m31.524s
>
> ## pg 12devel (cd3e2746)
> > time pgbench -i -s 100
> # done in 38.68 s
> # (drop tables 0.00 s, create tables 0.02 s, generate 29.70 s, vacuum 2.92 s, primary keys 6.04 s).
> real 0m38.695s
>
> That is an overall +20% regression, and about 40% on the generate phase
> alone. This is not a fluke, repeating the procedure shows similar results.
>
> Is it the same for other people out there, or is it only something related
> to my setup?
>
> What change could explain such a significant performance regression?

I think the pre-release packages have had assertions enabled at some
point. I suggest checking that. If it's not that, profiles would be
helpful.

Greetings,

Andres Freund