24 Jul
2006
24 Jul
'06
2:23 p.m.
Chuck Guzis wrote:
Hmmm... I always felt the 2901 constrained you rather too much, and that
some useful signals were not brought out. (OK, not as bad as the 2910
sequencer, which rapidly becomes a pain if you need more than 12 bits of
microcode address [1]).
[1] OK, I am biased. The PERQ 1 CPU used a 2910 and 4K of control store.
All later PERQs had 216K of control store, but kept the 2910 for
compatability. There was a circuit, commonly knwon as the '2 bit kludge'
(pun totally intentional) to extend the address from 12 bits to 14 bits.
And a kludge it is...
On 7/23/2006 at 11:52 PM woodelf wrote:
You didn't mention how "vintage"
your ALU is going to be (I'm assuming this
is a vintage design and not an FPGA). Are you using something like the
74S481 4-bit slice or are you going for something from the 29xx family?
I was planning to use generic 2901's and LS parts other than PROM. Ripple carry for the ALU is used since this is only 5
slices. Since I am
A 74S182 is only a 16 pin DIL package, and would give you lookahead
carry. Take a look at the PERQ 2 schematics on bitsavers -- the 16K CPU
board diagrams are there. And look at the PERQ CPU Techref also on
bitsavers. The PERQ is a 20 bit machine using 74S181 ALUs, and the carry
circuit should be applicable to you.
-tony