On March 30, brian wheeler wrote:
> > For 500 connections or more, I'd a brung in my electric wirewrap
> > GUN.
>
> A few years ago I took a hardware class at IU and we built a PDP-8 using
> PLAs, a bit of static ram, a few 74LS chips, and a ton of wire
> wrapping. There were about 600 wires in all, but luckily it was split
> across the whole semester, so it wasn't too dramatic. It was pretty
In 1985-1987 I worked at Princeton University (I think I may have
mentioned this at one point) on the Navier-Stokes Supercomputer
project at the Moody Fluid Dynamics Laboratory. I was responsible for
the Switching Board, a big board full of TTL that functioned as a big
crossbar switch between processors and memory planes. The boards in
the NSC machine were Unibus form-factor; we used the OEM version of
the PDP11/24 chassis which came with no labels and no backplane
wiring. The Switching Board was wall-to-wall 74LS series
chips...mostly 74LS244s...packed as closely as possible. If memory
serves they were 14,000 wires per board or so.
Doing that without a gun would have been...well, impossible. It was
tedious, but I enjoyed it anyway. What was especially enjoyable was
seeing the first test program running through vectors, outperforming a
uniprocessor Cray XMP on $2K worth of hardware in a desktop chassis. :)
-Dave
--
Dave McGuire "...it's leaving me this unpleasant,
St. Petersburg, FL damp feeling on my shorts..." -Sridhar
Received on Sat Mar 30 2002 - 15:37:13 GMT