22 Jun
2004
22 Jun
'04
10:44 a.m.
On 22 Jun 2004, at 08:53, Tony Duell wrote:
Interesting. The only machine I've ever come across that did something
like that weas the I2S (International Imaging Systems) Model 70 Image
processor/display. Those machines have several byteplanes, the outputs of
which feed lookup tables, the outputs of those are added (there's a large
board of full adders in the rack), then to more lookup tables. The manaul
suggests programming log and antilog tables if you want to multiply the
byteplanes.
Has anyone else ever come across these I2S machines?
Perhaps I should have said that all calculators _that I have worked on
and understood the internals of_ (which basically means quite a few HP
models) use a long multiplication algorithm.
-tony
There have been discussions over here about teaching maths in schools,
in that with the common use of calculators it's not necessary for
students to understand things like long multiplications (the fact
that's
_exactly_ how every calculator I've ever examined does multiplication
is
another matter, but anyway). To some extent that's true, but then you
should teach the correct use of calculators. Why some statements that
are
mathemetically correct are not suitable for machine calculation. The
problem with itterations that converge far too slowly, or are unstable.
Things like that.
Well I can think of one programmable calculator that I used in the
early 1970s
that did multiplication/division etc using logarithms. This had the neat