[vcf-midatlantic] Working on a historical microprocessor exhibt

Adam Michlin amichlin at swerlin.com
Sun Feb 9 14:58:15 EST 2020

On 2/9/2020 2:24 PM, Dave McGuire via vcf-midatlantic wrote:
>    Bit-slice is great stuff.  There are many bit-slice chips, but the
> Am2901 essentially owned the market.  It is a four-bit unit.
>    Basically it's a processor building block used to construct a
> micromachine.  You stack them up to whatever bit width you want, and
> program them with microcode to implement whatever macro-architecture you
> want.  There are companion chips, like the Am2910 sequencer, etc.  These
> are not required but come in handy.  The sequencer can implement
> constructs like a microprogram counter, branches and jumps, etc.
>    A few examples you've heard of: The VAX-11/730 uses a row of eight
> Am2901s to implement a 32-bit VAX processor, and the DECsystem-2020 uses
> a row of nine to implement a 36-bit PDP-10 processor.  Both use
> RAM-based control stores, so when they power up they don't speak the
> VAX/PDP-10 instruction sets until after the microcode is loaded.
>    This is generally the smallest/cheapest approach to build a large
> processor.  The VAX-11/730 and DECsystem-2020 are very small, very
> low-end members of their respective families.  But the bit-slice chips
> themselves are quite fast for their day, being big hot-running 1970s
> bipolar chips.
>    It's also possible to use Am2901s (or other bit-slice chips) as
> processors in their own right.  For example, many, many third-party disk
> controllers for DEC machines use a pair or a quad of Am2901s to
> implement their functionality.  In this application they typically
> aren't running microcode, but are used as more of a microcontroller with
> firmware.  The distinction is largely semantic, though, of course.
>    It's great stuff.  If you want to learn more about how all of this
> works, I recommend picking up a copy of "Bit-Slice Microprocessor
> Design" by John Mick and James Brick.  This book is/was known in the
> industry as "Mick & Brick".

That fills a huge hole in my understanding of architecture history. It 
certainly makes sense that there would be separate ICs that eventually 
merged into one discrete CPU. I just assumed it was done by the 
minicomputer manufacturers themselves. I had no idea it was AMD! Thanks!

>> I'd love to find a Zilog Z8000, but don't imagine period correct Z8000s
>> are so easy to come by. We have a Z8000 machine in the warehouse, but
>> one of the big rules for this product is that no vintage computers will
>> be harmed to make it happen.
>    Are you talking about just a Z8000 chip?  I can provide that for the
> exhibit.  It will need ESD protection, of course.

Cool, we'll talk. And not only ESD protection, but non degradable ESD 
protection is being planned.

More information about the vcf-midatlantic mailing list