Re: Improving btree performance through specializing by key shape, take 2

On Wed, 27 Jul 2022 at 13:34, Matthias van de Meent
<boekewurm+postgres(at)gmail(dot)com> wrote:
>
> On Wed, 27 Jul 2022 at 09:35, Matthias van de Meent
> <boekewurm+postgres(at)gmail(dot)com> wrote:
> >
> > On Mon, 4 Jul 2022 at 16:18, Matthias van de Meent
> > <boekewurm+postgres(at)gmail(dot)com> wrote:
> > >
> > > On Sun, 5 Jun 2022 at 21:12, Matthias van de Meent
> > > <boekewurm+postgres(at)gmail(dot)com> wrote:
> > > > While working on benchmarking the v2 patchset, I noticed no
> > > > improvement on reindex, which I attributed to forgetting to also
> > > > specialize comparetup_index_btree in tuplesorth.c. After adding the
> > > > specialization there as well (attached in v3), reindex performance
> > > > improved significantly too.
> > >
> > > PFA version 4 of this patchset. Changes:
> >
> > Version 5 now, which is identical to v4 except for bitrot fixes to
> > deal with f58d7073.
>
> ... and now v6 to deal with d0b193c0 and co.
>
> I probably should've waited a bit longer this morning and checked
> master before sending, but that's not how it went. Sorry for the
> noise.

Here's the dynamic prefix truncation patch on it's own (this was 0008).

I'll test the performance of this tomorrow, but at least it compiles
and passes check-world against HEAD @ 6e10631d. If performance doesn't
disappoint (isn't measurably worse in known workloads), this will be
the only patch in the patchset - specialization would then be dropped.
Else, tomorrow I'll post the remainder of the patchset that
specializes the nbtree functions on key shape.

Kind regards,

Matthias van de Meent.