On Thu, 2 Jun 2022 at 20:20, John Naylor <john.naylor@enterprisedb.com> wrote:
> If anything, changing work_mem is an
> easy to understand (although sometimes not practical) workaround.
I had a quick look at that for the problem case and we're very close
in terms of work_mem size to better performance. A work_mem of just
64.3MB brings the performance back to better than PG14.
postgres=# set work_mem = '64.2MB';
postgres=# \i bench.sql
c1 | c2 | c3 | c4 | c5 | c6
----+----+----+----+----+----
(0 rows)
Time: 28949.942 ms (00:28.950)
postgres=# set work_mem = '64.3MB';
postgres=# \i bench.sql
c1 | c2 | c3 | c4 | c5 | c6
----+----+----+----+----+----
(0 rows)
Time: 19759.552 ms (00:19.760)
David