Re: 32 Bit 1802
- It wasn't my intent to discourage anyone who wants to design a new
32-bit CPU from doing it but there are some real and standardized
choices already out there.
I have thought about building a multi-processor system like Richard
suggests for some time but haven't yet built up enough steam to put
the ideas into a real project. RCA made several nice little cmos
controller chips (besides the 1861) and one could build such a system
easily using them.
The RCA chips are a little difficult to find these days but other
chips such as the 6845 can still be easily found. The 6845 could be
driven by an 1802 using it's own memory space and communicating with
the main processor via a parallel port as Richard also suggests.
The same concept could be followed for other important major functions
such as a floppy or IDE I/O controller and even bit level I/O for
control functions all using thier own 1802s or other micros.
By preserving the 65k memory space of the main CPU(s) one leaves a lot
of possibilies open. Parallel processing anyone? It is easy to forget
just how much one can do with 65k or less of memory and I for one have
not yet written applications on 8-bit machines that have used it all up.
It's an entirely different story when you start to talk about object
oriented compilers and the like. It's not uncommon to see megabyte
sized compiles to perform fairly simple functions. That's when the
programming and hardware details get to be more work than fun.
> But If your main concern, is the graphic ability.
> I have a suggestion. Write Code To operate some graphics card.
> with the assumtion that it would be connected to the 1802 using 2 I/O
> ports, the High 8 bits would just be put in a buffer chip, and when
> the low 8 bits was writen it would then write to a 16 bit card.
> Most of the fancy graphics we see now days is not comeing from the
> main processor, they are comeing from the graphics processor. You
> would then have access to the graphics memory through the interface.
> You would then start by writeing your own graphics bios.
> emulation would then return to emulating real hardware.
> Higher memory could also be done with just I/0 ports, to switch banks.
> Anyway, back to earth. You don't see 32-bit ultra-low-power microsThe problem is that (CMOS Field Effect) transistors generate heat (which is
> because more bits means more power. To truly minimize power, you must
> minimize clock speed and the number of transistors switching at that
> speed. We've made another one of those monoculture decisions, and went
> for MAXIMUM speed and MAXIMUM number of transistors.
nothing more than dissipating power) every time they switch from one state
to another. And more transistors means more switching going on, which means
COSMAC stands for COmplementary MetAl-oxide Conductor (a bit of a forced
name if you ask me :)), which was RCA's way of saying their processor was
made with CMOD Field Effect Transistors (MOSFET's).
If you use the bipolar transistors (instead of CMOS/NMOS Field Effect
Transistors), you even have to keep the transistor 'in conduction' (this is
how it's called in Dutch). If not, you won't be able to switch from one
state to another fast enough. But a bipolar transistor in conduction also
uses power when it's NOT switching. Which makes it even worse, although
bipolar transistors are faster switches (if you keep them in conduction).
Cray's used to be made of Ga-As (Gallium Arsenide) bipolar transistors. The
computers were fast, but generated VAST amounts of heat and used as much
power as a small town. A Cray-3 uses 88.000 watts. It ran at 500MHz, which
is laughable these days :+). Today a P4 uses about 75 watts and runs at