Bug#444271: any progress on iop32x hard drive performance patch?
On Thu, Feb 21, 2008 at 05:48:30PM +0100, Martin Michlmayr wrote:
> * Gaudenz Steinlin <gaudenz@debian.org> [2008-02-21 17:10]:
> > Are there any news on the patch tbm mentioned for #444271? Anything I
> > could test or investigate? I would really love to have better hard disk
> > performance on my N2100. Is the patch from Intel available somewhere?
>
> Intel sent me new patches about a week or two ago. So far, they look
> good. My test kernel can be found at
> http://merkel.debian.org/~tbm/tmp/linux-image-2.6.24-1-iop32x_2.6.24-4iop1_arm.deb
I tested this kernel today with heavy IO load (backup over NFS and local
copying of several GB of data) and did not experience any data
corruption on my N2100.
The disk performance is approx. 2x better than before, but not yet as fast as tbm
measured on his box with the original Thecus firmware:
now on an idle system with 2.6.24-1-iop32x:
minotaurus:/home/gaudenz# hdparm -T -t /dev/sda /dev/sda1 /dev/sda2 /dev/sda3
/dev/sda:
Timing cached reads: 328 MB in 2.00 seconds = 163.97 MB/sec
Timing buffered disk reads: 122 MB in 3.03 seconds = 40.22 MB/sec
/dev/sda1:
Timing cached reads: 324 MB in 2.00 seconds = 161.61 MB/sec
Timing buffered disk reads: 22 MB in 3.05 seconds = 7.21 MB/sec
/dev/sda2:
Timing cached reads: 328 MB in 2.00 seconds = 163.92 MB/sec
Timing buffered disk reads: 14 MB in 3.08 seconds = 4.54 MB/sec
/dev/sda3:
Timing cached reads: 338 MB in 2.00 seconds = 169.12 MB/sec
Timing buffered disk reads: 16 MB in 3.31 seconds = 4.84 MB/sec
Strangely I measure much lower performance if I on the individual
partition:
On all my other systems the performance on the disk and the partitions
is about the same.
Gaudenz
--
Ever tried. Ever failed. No matter.
Try again. Fail again. Fail better.
~ Samuel Beckett ~
Reply to: