On Tue, 23 Mar 2004, Edouard Debry wrote:
> On Mon, Mar 22, 2004 at 12:44:09PM -0800, Mark Vojkovich wrote:
> > My impression was that "mem=nopentium" was only applicable
> > to older kernels.
>
>
> >Basically, the large page extension on AMD
> > processors behaves differently than on Intel. Linux kernel
> > did not properly support it (nor did Microsoft Windows, for that
> > matter). I thought that newer 2.4 kernels had workarounds for
> > the Athlon, but "mem=nopentium" was required to disable the broken
> > support altogether on older kernels. Perhaps this is a regression
> > in 2.6.
> >
>
> I also knew that "mem=nopentium" was deprecated with 2.6.xx kernels,
> do you know if this issue has been fixed in 2.6 kernels ??
I do not. I merely assume that if it was fixed in later 2.4
kernels, it would be fixed in 2.6 kernels.
> actually, I discovered that my memory was quite bad, at least
> one hundred errors with memtest !! so instabilities with AGP
> could be related not specificly to AGP software but to the fact that
> AGP uses one bad memory area
>
That sounds like a reasonable explanation.
Mark.
_______________________________________________
XFree86 mailing list
XFree86@xxxxxxxxxxx
http://XFree86.Org/mailman/listinfo/xfree86
