>IIRC, the NVAX5 chip is pin-compatible with the AXP 21064 chip. is
>this true? 

Largely.

>if so, my next question is, has anyone had any experiences with
>replacing the processor in an older-model Alpha box with an NVAX processor
>for a fast VAX? 

Yes, actually I have. I did it as a feasibility study a few years ago
for someone who wanted to explore the possibility of building a PCI-based
VAX. It is possible, but there are some not-so-obvious pitfalls and a
few compromises that must be made. Among them, for example, is that 
even though the NV5 is pin-compatible, the Alpha has a much larger
physical address range. The I/O is distributed throughout the address
range. You need to make certain the VAX can touch all the important
I/O areas, which is a bit tricky.

>Next, assuming this is possible, how on earth could one go
>about getting one's hands on an NVAX5 chip in this day and age??? is this
>a pipe dream? 

Yes, this is a pipe dream. DEC has manufactured all of the NV5s that are
ever going to be manufactured and has stockpiled them for their use. They
are unwilling to sell them and, when they do, the price is outrageous
(consider that every chip they sell you is one less high-margin VAX
they can build). It is simply not cost-effective.

>:) i wont even begin to worry about firmware for such a
>half-breed yet! 

I have a nice little 32-bit FORTH that loads into the NV5's cache from
the serial ROM then bit-bangs the serial ROM I/O leads to provide a
console. It is capable of living entirely within the cache, and is even
intelligent enough to move itself out of the way when it's time to test
a chunk of memory (a good thing too; I never would have figured out one
of the obscure pitfalls without that particular feature).

>Also, far less related, assuming i can't find an NVAX5 or put it into, for
>example, a DEC3000, is there a way to put a 21164 into one? 

Not as far as I know. I've never actually designed around the 21164 bus,
but I understand that it is quite different from the 21064/NV5 bus.

Roger Ivie
ivie@cc.usu.edu