> On 19 Aug 2025, at 15:46, Andres Freund <andres@anarazel.de> wrote:
>
> I think it's pretty insane to do things like variance computation while
> holding a spinlock, for every friggin query execution. The spinlock'ed section
> is ~185 instructions for me, with plenty high-latency instructions like
> divisions.
>
> It's so slow that it has measurable impact for single threaded readonly
> pgbench. Which does friggin btree lookups.
Do you think it could be an option to first add regression benchmarks,
and then consider migrating the most time-consuming columns
to another extension?
Not sure moving results out of pgss could be an option.
---
Benoit