Re: problems with postgresql speed - Mailing list pgsql-admin

From Jay Wren
Subject Re: problems with postgresql speed
Date
Msg-id Pine.BSO.4.33.0112041043230.20119-100000@pain.xmtp.net
Whole thread Raw
In response to Re: problems with postgresql speed  (Ludva Radomír (KM) <Radomir.Ludva@cssz.cz>)
List pgsql-admin
I know someone already pointed out the cache differences between the
processors, and that is likely to contribute to the differences you have
observed.

As was stated in the message about cache, databases are extremely IO
bound.  It is worth noting that Celeron's have a Front Side Bus speed of
66Mhz, while P3's have a FSB of 100Mhz or 133Mhz.  If you are running a
133MHz FSB, that is twice the data rate between your RAM and your CPU than
the Celeron.

Also, you seem to imply the systems are identical other than the CPU, but
Disk IO is nearly always the bottleneck in any system.  If one disk is
slower than the other, great differences in speed can be noticed even in
the same system.

Finally, my guess is that the P3's greater cache and faster memory bus
are what give over twice the performance on the P3.

-J

On Mon, 3 Dec 2001, [iso-8859-2] Ludva Radom�r (KM) wrote:

> I have got the same problem.
> I am working on it about a month :(.
> It is not hot :-)
>
> Two machines:
> 1. Pentium3 400Mhz, hdd 8Gb, 64Mb RAM, kernel 2.2.14-5.0 (RedHat 6.2)
> 2. Celeron  400Mhz, hdd 8Gb, 64Mb RAM, kernel 2.2.14-5.0 (RedHat 6.2)
>
> The same database, the same postgres Version 7.1.3, the same configuration.
> Only processor is diferent, ...
>
> Queryes from client application on Windows over ODBC:
> 1. 40sec.
> 2. 90sec.
>
> Radmi
>
> ---
> Odchoz� zpr�va neobsahuje viry.
> Zkontrolov�no antivirov�m syst�mem AVG (http://www.grisoft.cz).
> Verze: 6.0.303 / Virov� b�ze: 164 - datum vyd�n�: 24.11.2001
>
> ---------------------------(end of broadcast)---------------------------
> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org
>


pgsql-admin by date:

Previous
From: Tom Lane
Date:
Subject: Re: compression -Fx "problem"
Next
From: Shaun Thomas
Date:
Subject: Re: problems with postgresql speed