On Mon, Nov 30, 2015 at 11:00 PM, David Rowley
<david.rowley@2ndquadrant.com> wrote:
> There are in fact also two queries in TPC-H (Q10 and Q18) which are written
> to include all of the non-aggregated column in the GROUP BY list. During a
> recent test I witnessed a 50% gain in performance in Q10 by removing the
> unneeded columns from the GROUP BY clause.
Wow, that's pretty impressive. +1 for trying to do something about this.
(Full disclosure: I have not read your patch.)
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
