Hmmm... One 'thought experiemnt' goes
like this :
Take your favourite microprocessor and write a
for it. Burn that program into an EPROM and wire it up [...]. Most
people agree that EPROM cotnains 'firmware'.
But an EPROM is just a combinatorial logic
circuit. So for each of
the 8 data bits of the EPROM, write down the logic equation giving it
as a functionof the address inputs. Now implement that logic
equation using TTL gates. Repeat for the other EPROM data pins.
Wire the mess up, add tri-state buffers to the oputput, and connect
it in place of the EPROM. The microrpccor runs as before.
Well, to be picky, I'm inclined to doubt it. The pile of TTL (quite
aside from drawing enough power to heat your house) is very likely to
have substantially higher propagation delay than the original PROM.
I wonder... The older EPROMs had quite high access times (450ns was quite
common). I susepct I could make a 128 location 'TTL ROM' which was faster
than that without much effort.
It must do. It gets the same logic levels on its
data pins in the 2
cases for each address it outputs. The microprocessor cann't know
whether it's an EPROM or a mass of 74xx chips.
But the result sure doesn;t _look_ like firmware
I'd say it looks as much like firmware as a PROM implemented as a board
full of diodes, some of which have been cut. :-)
:-). I do find it amusing that the printset for the DEC diode matrix boot
ROM for the PDPD11 (M792?) includes schematics for the standard versions
showing only those diodes that are fitted. A scheamtic for waht is
essentialyl a piece of software is certainly odd...
For that matter, It's difficult to
distinguish between 'microcode'
and a 'state machine'.
In theory. In practice, there is usually little doubt - while the two
formalisms may be formally identical, there are, as reflected in the
terminological difference, differences of attitude and style between
Sure there are things that are clearly micrcode (like the control program
on the 09810-66513 baord in my hP9830). There are things that any
reasonable person would call state machiens (like the memeory timing
circuit in that same machine). But what about the disk controller (or
ethernet interface) in a PERQ? There's a ROM, sequenced by a 2910 chip.
It controls the various FIOFs, data encoders, CRC genrator, etc. Is that
microcode or a state machine. I think it's on the border ;-)