Extremely slow hard disk performance in FC3.

[Peter Teuben]

I filed a bug report on this (or seemingly this) a long long time ago
when it was a pure redhat problem. At the time I found that both Gnome
and KDE have this problem. I also found the solution, turn off
magicdev off.   I've documented this (and other problems on the 8200)
on 
	http://www.astro.umd.edu/~teuben/linux/laptop/dell8200.html
but i should update this and link the bugzilla ID in there. I haven't
recently checked what the status of this bug is....


- peter

On Mon, 15 Nov 2004, Ismael Juma wrote:

> Hi,
> 
> I have been experiencing some odd behaviour (major slowdowns) in my
> laptop for some weeks now. After some investigation it turned out that
> hard disk performance was the reason. My buffered reads were way below
> the average when running hdparm -tT /dev/hda. However, this only
> occurred at some points in time. Since it was hard to reproduce I did
> not report the problem.
> 
> Yesterday I have finally been able to narrow down the cause of the
> problem. This slowdown occurs if my cd-rom drive is open or if there is
> no disk in it. Even though I don't understand how this is possible, the
> figures from hdparm -tT /dev/hda follow:
> 
> With a cd in the cd-rom drive (average of three runs):
> /dev/hda:
>  Timing cached reads:   605.79 MB/sec
>  Timing buffered disk reads:   29.90 MB/sec
> 
> With the cd-rom drive open (average of three runs):
> /dev/hda:
>  Timing cached reads:  594.06 MB/sec
>  Timing buffered disk reads:   5.11 MB/sec
> 
> With cd-rom drive closed but with no cd in it (average of three runs):
> /dev/hda:
>  Timing cached reads:  593.94 MB/sec
>  Timing buffered disk reads:   14.7 MB/sec
> 
> My system is a Dell Inspiron 8200 laptop (P4M 1.9 Ghz, 1 GB RAM, Geforce
> 4 Go 64 MB RAM) and /dev/hda is a Hitachi Travelstar 7k60 (7200 RPM
> drive). I also have an external firewire drive that does not exhibit any
> of the problems mentioned.
> 
> I'm using a fully updated Fedora Core 3 with a 2.6.10-rc2 kernel and KDE
> 3.3.1. I have experienced this problem with earlier kernels though. It
> is important to note that the current problem is _always_ reproducible.
> 
> Has anyone experience a similar problem? Should I file a bug report and
> if yes, against what component?
> 
> Thanks in advance.
> 
> Regards,
> Ismael
> 
> P.S. DMA is enabled.
> 
>