<He said _a_ 2909. That implies a 4 bit microcode address, which limits it
<to 16 locations. 16 micorocode words and a 4 bit ALU is a small processor
<IMHO. And being an AMD design they won't have simply concatenated an
<'instruction' from external memory (which this board probably doesn't
<have anyway) with those 4 bits.
The 16word microcode limit is the greater problem but, once you understand
how it word concatinating a few more solves that nicely. It's one heck of a
demonstrator.
<For people who've not come across the AMD sequencer chips, they're quite
<nice, and even have things like a return stack (about 4 levels IIRC) to
<allow for subroutines in the microcode.
Depends on which one. the 2910 is 5 and the 09/11 is 4. They are pretty
useful parts for lots of things.
<The 2910 is essentially 3 2909s together with some glue logic in a single
<chip. If you only need 4K of control store it's a nice chip _but it can't
<be extended to give more bits without kludging_. Three rivers made that
<mistake on the PERQ -- the first PERQ had 4K of control store and used a
<2910, all later models had 16K and used a 2910 and a '2 bit kludge' (pun
<intended!).
;) 2910 is a good part and easy to live with unless your doing complex
stuff.
Allison
Received on Sun Feb 27 2000 - 18:26:23 GMT