On Tue, May 09, 2006 at 11:36:51PM +0200, Tonnerre LOMBARD wrote:
> Salut,
> 
> On Tue, May 09, 2006 at 12:06:42PM +0200, Manuel Bouyer wrote:
> > > I added the O_SYNC parameter to your open() and it sanitized the output
> > > quite a bit:
> > 
> > As the tool is using the raw device it shouldn't matter. What problem did
> > you notice ?
> 
> The read speed was around 4000MB/s, which was unreasonable.

Hum, then I guess you used the block device instead of the raw device

> 
> > BTW, you could also try to benchmark raw write speed: boot single user and
> > use:
> > dd if=/dev/zero of=/dev/rwd0b bs=64k (assuming your swap partition is wd0b)
> 
> There appears to be a significant problem in write speed:
> 
> vic# dd if=/dev/zero of=/dev/rwd0b bs=64k count=1536
> 100663296 bytes transferred in 41.410 secs (2430893 bytes/sec)

That's way to slow. So it looks like the issue is really somewhere
between the disk and the controller.
I would blame the controller, or more probably the way we use it (but without
doc it's hard to fix this :/

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--