Message-ID: <46E11ED7.3020708@aeccom.com>
Date: Fri, 07 Sep 2007 11:50:15 +0200
From: Sven Geisler <sgeisler@aeccom.com>
Organization: AEC/communications GmbH
User-Agent: Thunderbird 2.0.0.6 (Windows/20070728)
MIME-Version: 1.0
To: Thomas Finneid <tfinneid@student.matnat.uio.no>
CC: pgsql-performance@postgresql.org
Subject: Re: utilising multi-cpu/core machines?
References: <46DF3082.7040301@ifi.uio.no>
In-Reply-To: <46DF3082.7040301@ifi.uio.no>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

Hi Thomas,

PostgreSQL does scale up very well. But you have to keep in mind that
this also depends on profile of the application you're on PostgreSQL.

Insufficient memory and slow disk systems can interfere PostgreSQL.
Another issue is contention if the server has more than 4 cpus.
(Please check out discussions about "context strom" in this group.)

Anyhow, I had create a benchmark for my company which shows the scale up
of PostgreSQL 8.1.4. This benchmark does try to enforce contention
because of the profile of our application.

Clients/scale-up factor
1	1
2	1,78
3	2,47
4	3,12
5	3,62
6	4,23
7	4,35
8	4,79
9	5,05
10	5,17

Scale-up factor is relative to one client the number of completed
queries in a time frame. (throughput)

This test was done on a 16 core Intel-box (4-way Xeon E7340).

The results of TPC-B benchmark are looking similar.

Sven.


Thomas Finneid schrieb:
> Hi
> 
> I couldnt find any specifics on this subject in the documentation, so I
> thought I'd ask the group.
> 
> how does pg utilise multi cpus/cores, i.e. does it use more than one
> core? and possibly, how, are there any documentation about this.
> 
> thomas
> 
> ---------------------------(end of broadcast)---------------------------
> TIP 6: explain analyze is your friend

-- 
Sven Geisler <sgeisler@aeccom.com>   Tel +49.30.921017.81  Fax .50
Senior Developer, AEC/communications GmbH & Co. KG Berlin, Germany