🚀 go-pugleaf

>From article <64...@lindy.Stanford.EDU>,
by ri...@hanauma.stanford.edu (Richard Ottolini):
> There are probably some good stories about how so much US space
> exploration was managed with such primitive computers.
>    ...
> Part of the explanation is that it takes over a decade from design to
> deployment of a space experiment.

Here's an unfinished story:  The physics department at the U of Iowa is
currently building an experiment they intend to fly on the ISTP Polar
mission in a year or two.  (Note, these are the same people who built
Explorer I.)  I was involved in doing some of the preliminary algorithmic
design for data compression routines they want to use so they can get
more image data down to Earth, and here's what they've got:

A pair of 3.5 Mhz 80C86 processors, radiation hardened, for about $15,000
each.  Wait!  Yes I know this is a $5 part, but it takes about a decade
between the time a microprocessor hits the market and the time the same
architecture is available in low-power radiation-hardened form certified
for space launch.  We are only 20 miles from the plant where the Rockwell
AAMP microprocessor was designed, and I know it would fit our needs
perfectly, but if you're not part of Rockwell, just try to get your hands
on any software support products for that chip!

What about software tools?  We expect to push the limits of what our
nearly brain-dead processors can handle, so all my feasibility studies
have been done with hand-crafted assembly code.  Some of the optimizations
I've used have been the kind that no known compiler can do for me, and
they've given speedups of about 2 over more usual assembly coding
techniques.
				Doug Jones
				jo...@herky.cs.uiowa.edu