Re: SuSE 9.1 Pro performance problem
From: Vlad (miles123_at_gmx.de)
Date: 05/19/04
- Next message: Roland Schulte-Osthoff: "Re: Suse 9.1 kills Windows partitions. Use Linux-->Lose Data......."
- Previous message: Peter Tröhlmann: "Re: OpenGL enabled screensavers that came with KDE don't work"
- In reply to: -linux_lad: "SuSE 9.1 Pro performance problem"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: 19 May 2004 12:58:15 -0700
-linux_lad <john@linuxlad.org> wrote in message news:<zZCdnWW_VuBf6TbdRVn2hw@giganews.com>...
> Platform and hardware description:
>
> Athlon 3200 64 bit processor
> Gigabyte GA-K8VT800M motherboard
> 512MB PC2700 333MHz DDR Certified
> Maxtor 80 GB SATA with 8 MB cache
> Dynatron k8 fan
> Pristine SuSE Pro 9.1 64 installation with all available patches except
> nvidia
>
> Problem
> The system seems to be running somewhat slower than other comparable
> systems. I know serveral people with almost identical configurations,
> and they report better results on their machines with the bechmark
> script than my machine, running the same script. I work as an Oracle
> DBA, so I have access to 8-way and 4-way sparc and RS-6000 machines.
> I also have access to a couple of dedicated Opteron 140 machines, one
> running FreeBSD and the other Running 9.1 Pro and they both generate
> comparable results. On all of the platforms I have run the script on,
> the results have been predictable, but the machine enumerated above
> seems to be running about nineteen or twenty percent slower than other
> machines in similar configuration, OS and processor power. I'll post
> this in the other suse group too.
>
Hello John,
unfortunately there aren't enough information to exactly determine the
reason why the one machine is slower.
It appears that your tests are prity I/O intensive. So it may
be that there is an I/O bottleneck. I suggest that parallel to
the test you start iostat and monitor the %iowait parameter.
When you say "similar systems" does it mean that they also have
the same hard-disk and that they have the same file-system and
block size of the file system ? Block size does have influence
and also whether the fs has logging or not.
Did you perform the test on a new formated filesystem ? I am
asking that, since fragmentation can have influence on the I/O
performance.
Regards,
Vlad!
- Next message: Roland Schulte-Osthoff: "Re: Suse 9.1 kills Windows partitions. Use Linux-->Lose Data......."
- Previous message: Peter Tröhlmann: "Re: OpenGL enabled screensavers that came with KDE don't work"
- In reply to: -linux_lad: "SuSE 9.1 Pro performance problem"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
