Re: index prefetching

On Thu, Aug 14, 2025 at 7:26 PM Tomas Vondra <tomas(at)vondra(dot)me> wrote:
> Good. I admit I lost track of which the various regressions may affect
> existing plans, and which are specific to the prefetch patch.

As far as I know, we only have the following unambiguous performance
regressions (that clearly need to be fixed):

1. This issue.

2. There's about a 3% loss of throughput on pgbench SELECT. This isn't
surprising at all; it would be a near-miracle if this kind of
prototype quality code didn't at least have a small regression here
(it's not like we've even started to worry about small fixed costs for
simple selective queries just yet). This will need to be fixed, but
it's fairly far down the priority list right now.

I feel that we're still very much at the stage where it makes sense to
just fix the most prominent performance issue, and then reevaluate.
Repeating that process iteratively. It's quite likely that there are
more performance issues/bugs that we don't yet know about. IMV it
doesn't make sense to closely track individual queries that have only
been moderately regressed.

--
Peter Geoghegan