--- Begin Message ---
James Pollard wrote:
> Quoting [EMAIL PROTECTED] ([EMAIL PROTECTED]):
> > Thanks all for your input.
> >
> > The problem is solved in a way.
> >
> > Before mucking with hardware i tried few software things.
> >
> > With 2.0.36 compiled as 686 and as 586 i had only 350Bogomips with a 350MHz
> > K6-2
> >
> > BUT (!) with 2.2.0pre7 compiled with CONFIG_M586TSC=y i have my full 700
> > bogomips !
> > with it /proc/cpuinfo shows everything right (CPU clock,etc...)
> >
> > That's strange isn't it ? Probably a strange compilation quirk somewhere....
> >
> > I'll stick with 2.0.36 for now (as 350Bogomips should be enough to serve
> > only
> > few web pages), but i will wait anxiously to 2.2.0 final
> >
> > Patrick
> >
> >
> > --
> > Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] < /dev/null
The speed-up might be caused by an improved hardware-level loop-function which
has been
used in time-sensitive code. Maybe the new kernel handles tha piece wrongly.
The problem
is known that i.e Borland Pascal programs crash on any AMD K6 =>350Mhz and that
the
mouse pointer get weird in Win98 due to this incompatible timer fucntion.
something might be found on heise.de in the c't archive (or current
heise-register).
Btw.: Has there BEEN any speed-up? I don't think so !?
--- End Message ---
