[Info-vax] Request for AlphaServer 1000A SIMM/SIMMS

[John Wallace]

On Jul 5, 10:41 pm, Steven Schweda <sms.antin... at gmail.com> wrote:
> On Jul 5, 1:43 pm, urbancamo <m... at wickensonline.co.uk> wrote:
>
> > All the documentation for the AS1000A [...]
>
>    Which "All", where?  I know nothing about the 1000A, but a
> little poking around at
>      http://h18002.www1.hp.com/alphaserver/archive/
>      http://h18002.www1.hp.com/alphaserver/archive/1000a/
> suggests that things might depend on whether the CPU type is
> EV4 or EV5 (which "AlphaServer 1000A", without its model
> suffix, fails to reveal).  And there are different "Supported
> Options List" documents for different 1000A system types.
> (Message-shortening techniques, like omitting useless
> information, are best left to those who know which information
> is useless.)
>
> > [...] also listed as being for an AlphaServer 200 4/233.
>
>    It was "also listed" where?  I believe that the memory in
> an AlpSta 200 4/233 was plain FPM, not EDO.  The "200 Series"
> book doesn't mention EDO, only (true) parity and 60ns or 70ns.
> (The book also doesn't mention that 128MB modules will work,
> but they do.)  I would not bet that EDO modules would work in
> a 200, and I would tend not to rely on a source which says
> that they do.

There's another cautionary point to note about some of the 21064 based
AlphaStation boxes, and that is that they used parity memory, as you
mentioned, but with a bit of a twist.

Early in the lifetime of the product, 36bit memory (ie capable of full
ECC) was sometimes used, but later on, it was "value engineered" down
to 33bit (no byte access, so only one parity bit needed per 32bits),
which won't do full ECC. Some customer part numbers didn't change even
though the parts were not functionally the same.

Care is needed.