[Info-vax] Another Alphaserver 1200 memory question

[H Vlems]

On 29 jan, 04:14, moro... at world.std.spaamtrap.com (Michael Moroney)
wrote:
> H Vlems <hvl... at freenet.de> writes:
> >According to the Systems&Options guide the AlphaServer 1200 had three
> >memory options: MS300-BA (64 MB), MS300-DA (256 MB) and the MS300-EA
> >(512 MB).
> >Now the -CA code is missing and I wondered whether two 64 MB boards
> >would make a "mythical" MS300-CA.
> >When added, SRM reported the two boards as a 32 MB memory option and
> >that is what VMS also reported.
> >I had expected to see the experiment fail, unable to complete T24, or
> >unable to boot, or, since part of the console lives in upper memory, a
> >spectacular crash when VMS boots.
> >Who can supply an explanation for displaying only 25% of the installed
> >memory pair?
>
> Is what you have a "real" MS300-CA or just two 64 MB boards?  (does the
> -CA exist?)
>
> The console code has to know of supported/possible memory combinations
> so it can map them in place and locate the in-memory portion of the
> console code.  It may not know of the 128MB combo because DEC never
> planned to sell the -CA for it, and perhaps it just mapped it as a
> default which happened to be 32MB.  (perhaps a -AA was planned, for NT
> or something)

No, an MS300-CA does NOT exist AFAIK. I just made that up because it
happens to be a missing suffix :-)
The thing is, a 16 MB memory pair also never existed. So it seems that
the console firmware tries to interpret the
memory that is installed, does a few checks (same size bavks,
registered memory and possibly a speed check) and
tries to figure out what capacity is available and works with the
computed result.
Using just the two 64 MB boards does not work: the POST reports a
console reset error, possibly because the console components that live
in the lowest memory pages and the highest memory pages together are
too sizeable to leave enough memory to fit VMS in.
Hans