> > I'm not quite sure when this problem occured first, but I
> > just saw that 1.6Z uses UDMA33 for my harddisk while it=20
> > detects UDMA100. I've tried to configure the kernel to only=20
> > use UDMA100, which would not work either. I'm using a SiS 650=20
> > IDE controller with a Fujitsu MHR2020AT. I don't think that=20
> > one is fast enough to actually gain performance by UDMA100=20
> > mode, but I like to see everything running smoothly. I'd be=20
> > glad if anyone had a suggestion.
>=20
> I'm not sure this is the problem here, but UDMA higher than=20
> UDMA33(66?) require a 80 pin UDMA cable. If you try to run=20
> UDMA100 on a "normal" IDE cable, the kernel will downgrade to=20
> the next lower transfer mode.

No, some older kernels and the 1.6.1-INSTALL precompiled kernel actually
used (or at least told they do) UDMA100. Current precompiled current- =
and
1-6-release- won't. May there be any traps in kernel configuration which =
may
lead to such results?