mainframes and other stuff

[William Donzelli]

> I'm VERY skeptical of this one!  First, I know IBM went to great lengths at
> some time in
> the early 60's to figure out how to make reliable computers with a LOT less
> gold than
> the traditional wisdom required.  As far as I know, all 360's had a LOT less
> gold in
> them than similar-sized machines from other makers at the same time.

That is not true. Early S/360 had a lot of gold in the backplane pins,
later ones did not. This is why IBM stuff from 1970 or so is such a
mixed bag - some machines had the cost saving tinned backplanes, some
still had the gold.

Looking at one of the gold backplanes, you will see that the plating
is thick - the gold is very deep and dull. Over that many pins, that
is a very large surface area. Doing a thick plate over a large
effective surface area takes a LOT of metal. To give a more modern
perspective on things - how much gold do you think is in a Pentium Pro
microprocessor? Thick plate over a large surface area...go look it up,
and you will probably be surprised. Then do some back of the napkin
calculations on the surface area of the PPro, compared to thousands
upon thousands of pins in a 2075...

--
Will