6 Jul
2007
6 Jul
'07
6:49 a.m.
From: Jules Richardson <julesrichardsonuk at
yahoo.co.uk>
I've been thinking more and more this last week about designing a CPU to be
built from TTL logic ICs, purely as an interesting exercise.
I'm thinking of a microprogrammed design with a pretty minimal register
count - but as speed is never going to be a defining aspect anyway using
commodity TTL, I'm thinking of going for a bit-serial ALU to keep the parts
count down. Keeping the raw components pretty basic is another desired goal
- no custom off-the-shelf ALU chips, gate arrays etc.
---snip---
Hi
I've been thinking on a similar idea, only I'm not thinking of a traditional
ALU. I've been looking at how to create a minimal computer. There is
a web page that describes a processor with only one instruction.
I have been thinking of how to make it even simpler. First, they
us a subtract. I think that is overly complicated od an operation.
It requires adders connected with carries. This is always a clock limiter
when the add operation it covers is not needed for most code.
I believe just a NAND operation and branch on zero is sufficient.
Things like carry can be handled in tables.
The address counter bothers me as well. An incrementer is also
a pain. An LFSR generator is much simpler. It does require that
on have the ful complement of RAM because of the randomish
nature of incrementing. Still, the ciircuits would be much simpler
to implement. Coding is a pain and a method to bootstrap the
RAM with the tables would also be needed.
Dwight
_________________________________________________________________
http://imagine-windowslive.com/hotmail/?locale=en-us&ocid=TXT_TAGHM_mig…