FWIW, I've been working and benchmarking this a lot over the last weeks.
I'm running a lot of tests as weel, on HDDs. It is basically always better with the patch, although sometimes not sorting but flushing is better than both, which suggest that the gucs should be kept just in case.
After finally nailing down the performance regression due to wal writer, thins are looking good. I plan to post an updated version soon.
Good. The last version sent (14?) does not apply cleanly. I'm looking forward to have another look at an updated version.
-- Fabien. -- Sent via pgsql-hackers mailing list (firstname.lastname@example.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers