[Info-vax] DEC Large Systems

[Michael Moroney]

Bill Gunshannon <bill.gunshannon at gmail.com> writes:

>For a man as brilliant as Gordon Bell was he could sure be stupid
>sometimes.  32bit/36bit?  The user world probably never cared.  It
>was (and is) the applications that make a system worthwhile.  The
>PDP-11 started at 16bits, went to 18bits and eventually to 24bits.

22 bits, actually.

A big difference was that the PDP-11 was still 16 bits, despite the larger
address space. It could not access anywhere directly, but only through
windows into that address space.

>And I am sure it could have moved on from there to 32bits and even, 
>eventually to 64bits.  I suspect the 10 and 20 likewise.

It could, in fact the 8 bit 8080 lives on as the 64 bit x86, which is doing
quite well.  The VAX was developed in response to the PDP-11 issues, as a
rather different path (design rather than evolution).

Another issue was the high cost of developing new chips such as the next generations
of Alpha. You need a large market to support it. I remember someone from within
DEC calling that development the "Alpha cash furnace". But such development was
necessary since all other chip designers were improving their own products.

>And, comparing the any of this to the move from VAX to Alpha is
>really apples and oranges.  Word Size was trivial compared to the
>real architectural changes in that  move.  Wether or not the
>move to RISC was worth it is for another argument.  I expect that
>for most of the VAX customers at that time it was meaningless
>and really brought nothing of value to the table.

Speed.