Re: life extension of old computers thanks to opensource Contiki

[zwsdotcom@gmail.com]

> I think that 8-bit processors are not really interesting anymore. The only
> reason to use an 8-bit processor might be that the systems and the processor

I happen to be writing a book (my third, for what that's worth) about
how to become an embedded engineer. Here are a couple of quotes for
you:

--- 8< ---

Humor aside, it is a generally-accepted fact that the number of new
graduate engineers of all types is shrinking (at least in the United
States). Various theories are posited to explain this phenomenon. In
the specific case of embedded engineering, I see several factors
causing the decline. One such factor is the unavailability in this day
and age of well-documented home computers of the style my generation
enjoyed - Acorn BBC, Commodore VIC-20, Sinclair ZX Spectrum, and so
on. Modern personal computers are black boxes designed to run
shrinkwrapped software. They are shipped without programming tools and
with no technical documentation whatsoever. Operating system vendors
are actively working to confine software development to an exclusive
club of licensees, and lower-level development at the
direct-hardware-access layer is at best very difficult due to the
unavailability of chipset documentation (in many cases due to the
manufacturer's contractual obligations to preserve trade secret
information related to intellectual property protection mechanisms),
and the utter complexity and heterogeneity of PC hardware. In the good
old days, we could develop a homebrew program on our Commodore 64,
fine-tune it to the last instruction cycle, and show it to other people
with pride. This type of skillset is a vital component of embedded
engineering.

There are other factors that raise the bar to becoming an embedded
engineer, too - and I'll deal with them in the appropriate sections
of this book - but the point I'm making here is that it's simply more
difficult these days for kids to experiment with what they have at
home.

--- 8< ---

Intel's 8051 architecture is an industry-standard platform referenced
in almost every engineering course. When spoken of in polite company
(and even in books like this one), you'll generally see the number 8051
prefixed with the adjective "venerable". Personally, I would rather
describe it as "decrepit", but it's a solid fact that the 8051 is
still the world's best-selling microprocessor core. Barring some major
upheaval, this is likely to remain true for as long as the world is
still manufacturing and using 8-bit micros(1).

(1) - This is not quite equal to saying "forever". As die sizes
shrink, the packaging and bonding costs (bonding wires from the silicon
to pins that go out to the real world) start to dominate the cost of
the device. Raw 8-bit dice will always be cheaper than raw 32-bit dice
due to their lower transistor count, but by the time the chips are
packaged for sale, the price difference may be almost invisible. For
some applications, it is foreseeable that there will be no real cost
benefit to be realized by staying with an 8-bit architecture.

--- 8< ---

(end quotes)

As a learning platform, the 6502 is not a bad choice - it's actually a
fairly neat architecture, and there are parts based on the core still
in production. Having to work with the realtime side of a moving raster
is excellent experience for embedded design.

The 68000 (and 010) are extraordinarily inefficient CPUs by modern
standards, and complex to use and interface. There is little practical
value in learning the M68K instruction set because it's practically a
dead family.

If you intend to move up from 8 bits, it makes no sense to bother with
16 bits, you should go directly to a single-chip 32-bit ARM device.