Wouter Coene wrote:
>
> According to Timothy J. Massey ([EMAIL PROTECTED]):
> > On Sat, 28 Aug 1999 13:02:51 +0200, Wouter Coene wrote:
> > >They're not actually virtualizing it. The chip simply contains silicon for
> > >both IA64 and IA32, and full hardware access is available for both. I dunno
> > >how they sorted out the protected mode stuff though. Perhaps you can't run
> > >IA32-PM code with an IA64-PM OS (PM of course means Protected Mode).
> >
> > I'm not doubting your informtion, but where did you see this? I haven't
> > heard anything about this, and the original IA-64 specs specifically
> > mentioned that they would *not* be doing that. But, it's not like Intel
> > can't change their mind...
>
> I've got the information from a dutch computer magazine called
> Computer!Totaal. The actual sentence is "To maintain 100% compatibility with
> the old IA32 architecture, Intel is adding an extra IA32 part to the
> Merced".
>From what I read in Intel's guide, there are mechanisms in
place to switch back and forth from IA32/64 modes, just like
a v8086 task would time-share with other protected mode
tasks.
But the IA32 support looks like it's built in to the hardware.
BTW, the Intel docs claim the IA32 support is that of a PentiumIII.
However, system support, such as paging, is vectored through the IA-64
mechanisms. This is similar to how v8086 support
vectors a real mode program to protected mode features.
-Kevin
