[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-sparc
Subject:    Re: Bad disk io performance with RH 5.1, kernel 2.0.35
From:       Hyong Lee <hyongel () charon ! astro ! nwu ! edu>
Date:       1998-08-31 18:38:22
[Download RAW message or body]


The bonnie results are suspect because your 10 MB test can fit completely
in RAM.  Don't know what iozone does, though.  Also, your (probably bogus)
bonnie results show Redhat block reads at 2x SunOS and writes at half
SunOS, while iozone shows, as you say, reads for Linux at half the speed
of SunOS.  You should try a larger test size (the bonnie readme says at
least twice the size of physical RAM).

On Mon, 31 Aug 1998, Klaus Kaempf wrote:

> After changing from SunOS 4.1.2 to RedHat Linux 5.1 my Sparc 1 (26 MB RAM)
> is very slow. Running IO performance tests (bonnie and iozone) confirms
> this.
> 
> SunOS 4.1.2 on ufs partition:
> 
>               -------Sequential Output-------- ---Sequential
> Input-- --Random--
>               -Per Char- --Block--- -Rewrite-- -Per
> Char- --Block--- --Seeks---
> Machine    MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec
> %CPU
>            10   186 98.4  2105 76.3   994 69.8   136 98.6  2380 90.2 210.3
> 67.4
> 
> Writing the 10 Megabyte file, 'iozone.tmp'...21.757300 seconds
> Reading the file...8.082830 seconds
> 
> IOZONE performance measurements:
>         481942 bytes/second for writing the file
>         1297288 bytes/second for reading the file
> 
> 
> RedHat Linux 5.1, Kernel 2.0.35 (rpm file grabbed from ftp.redhat.com) on
> ext2fs partion on the same disk
> 
>               -------Sequential Output-------- ---Sequential
> Input-- --Random--
>               -Per Char- --Block--- -Rewrite-- -Per
> Char- --Block--- --Seeks---
> Machine    MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec
> %CPU
>            10   149 95.8  1149 82.2   362 78.8    94 93.8  5143 98.0 195.6
> 93.6
> 
> Writing the 10 Megabyte file, 'iozone2.tmp'...20.370000 seconds
> Reading the file...24.230000 seconds
> 
> IOZONE performance measurements:
>         514764 bytes/second for writing the file
>         432759 bytes/second for reading the file
> 
> 
> So read performance is about 2-3 x *SLOWER* on Linux.
> 
> Anyone with a good explanation ?
> 
> Klaus
> 
> -
> To unsubscribe from this list: send the line "unsubscribe sparclinux" in
> the body of the message to majordomo@vger.rutgers.edu
> 

-
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of the message to majordomo@vger.rutgers.edu

[prev in list] [next in list] [prev in thread] [next in thread]