On Tue, 2008-08-12 at 21:35 +0200, Loïc Minier wrote:
> Here are some performance tests I did with various arm kernels. Test
> was to hdparm -t /dev/sda, then sdb, twice; results below are average
> speed of each 4 hdparm runs sequence. The speed of sda versus sdb was
> always very close (the disk parts match).
>
> Kernel Version Read speed (MB/sec)
> ----------------------------------------------------------------
> 2.6.18-3-iop32x 2.6.18-7 21.46 # etch?
> 2.6.18-6-iop32x 2.6.18.dfsg.1-22 21.37 # etch.3
> 2.6.24-etchnhalf.1-iop32x 2.6.24-6~etchnhalf.4 16.22 # etch 1/2
> 2.6.25-2-iop32x 2.6.25-2iop1 20.84 # Martin's
> 2.6.25-2-iop32x 2.6.25-7 15.73 # lenny
> 2.6.26-1-iop32x 2.6.26-1 16.05 # unstable
Weird, I see 33MB/sec with Martin's kernel.
Ross
--
Ross Burton mail: ross@burtonini.com
jabber: ross@burtonini.com
www: http://burtonini.com
Attachment:
signature.asc
Description: This is a digitally signed message part