9 Mar
2011
9 Mar
'11
7:31 a.m.
Subject: Microcode .vs. Hardware (Was: hardware
multiply/divide functionality in CPUs)
Date: Wed, 9 Mar 2011 07:03:08 -0800
From: rickb at bensene.com
To: cctalk at classiccmp.org
Christian wrote:
Hi
If anyone comes across a Diehl Combiton or a SCM marchant Cogito and
doesn't want it laying around, I'd love to have one.
I fiddle with on when I was working for the University of Miami.
I used it to run data from paleotemperature stuff and C14 dating
stuff.
I never had any external storage. I can recall if it was tape of punch
cards.
It was all hand wired transistors as I recall. All packet into the box.
It had two delay lines, one for data and the other for program.
I recall cleaning the photo cell that was used on powerup. It had
a two hole metal tape. One the clock and the other the data.
I never did figure out how it coordinated the delay line with
the tape. The tape didn't seem to have any speed control.
Dwight
Dwight
I'd say no, microcode is software...
the Diehl Combitron (designed around 1965). The last one is
particularly
interesting as its word length is 55 bits but the
number of
significant
result bits is variable, depending on the given
number of R delay line
cycles. Multiplication is shift+add, division is shift and
substract/add
(depending on the sign of the previous remainder). This all is done in
HW
with just a hand full of flip-flops (in the whole processor!).
I thought that the Combitron was largely a microcoded machine, with the
microcode read into a delay line from a punched metal tape at startup.
The Combitron was designed by Stanley Frankel, who also designed the
LGP-30, the SCM/Marchant Cogito calculator, and a number of other small
computing devices. I had a phone conversation with Allan Frankel, son
of Stanley, and learned a lot about the machine and its history. Allan
helped his father construct the prototype of what became the Combitron
in their home. Based on what Allan explained to me, the operation of
the machine was controlled pretty much entirely by the "firmware" loaded
in at power-up when the tape is read in. The low count of flip flops
would follow in such a design, as a small number of flip flops are only
needed to store state information, and a few for holding temporary
microcode operation code information, and a few more for buffering data
circulating through the delay line(s).
There are more suitable examples of hardware multiply/divide (and square
root) in some other early electronic calculators, though some that might
be thought to have hardware implementations are actually microcoded in a
sense. For example, the Sharp Compet 20 uses a sequencer with a diode
ROM with feedback to control its operation.
Early calculators from Friden, Casio, IME and Canon tended to be hard
wired designs that implemented the multiply/divide logic with a lot of
gating and various simple state machines.
Rick Bensene
The Old Calculator Museum
http://oldcalculatormuseum.com