Re: Problem with grouping, uses Sort and GroupAggregate, HashAggregate is better(?)

From: Rolf Østvik (HA/EXA) <rolf(dot)ostvik(at)ericsson(dot)com>
To: "Dave Dutcher" <dave(at)tridecap(dot)com>, <pgsql-performance(at)postgresql(dot)org>
Subject: Re: Problem with grouping, uses Sort and GroupAggregate, HashAggregate is better(?)
Date: 2007-01-15 08:58:25
Message-ID: D7E5DB1F5F917E45AA4DF9F9F1306BEE31E2EC@esealmw111
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

> -----Original Message-----
> From: Dave Dutcher [mailto:dave(at)tridecap(dot)com]
> Sent: Sunday, January 14, 2007 5:12 PM
> To: Rolf Østvik (HA/EXA); pgsql-performance(at)postgresql(dot)org
> Subject: RE: [PERFORM] Problem with grouping, uses Sort and
> GroupAggregate, HashAggregate is better(?)
>
> > -----Original Message-----
> > From: pgsql-performance-owner(at)postgresql(dot)org
> > [mailto:pgsql-performance-owner(at)postgresql(dot)org] On Behalf Of
> > Rolf Østvik (HA/EXA)
>
> Have you tried "set enable_sort=off" with 8.1.2? I'm not
> sure if that will
> change anything because it has to do at least one sort. Its
> just a lots
> faster to do a hashagg + small sort than one big sort in this
> case. (I
> wonder if there should be enable_groupagg?)

Did you mean enable_sort = 'off' for 8.2.1?

I tried to set enable_sort = 'off' for both the
8.1.4 server and the 8.2.1 server.
Both servers used the same plan as Run 4 and Run 3 respectively.
There were of course some changes in the planner cost for the sort
steps, but the execution times was of course the same.

Regards
Rolf Østvik

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message Florian Weimer 2007-01-15 10:16:36 pg_trgm performance
Previous Message Luke Lonergan 2007-01-15 07:42:34 Re: max() versus order/limit (WAS: High update