Open source chip design

[kriesel]

Originally Posted by R.D. Silverman

The clock rate was slow [only 10 KHz] because the board was built with prototype wirewrap. [proof of concept]

Astonishingly slow. I knew folks who hand wire wrapped custom designed full size Unibus boards for VAX systems populated with ECL chips, ~1984, to do specialized image processing and real-time machine control at 2 microsec or better, far faster than the VAX 11/750 computer could process by itself. Those boards were the area of a closed pizza box and ran at 30Mhz IIRC. (It controlled a 3kw electron beam making little raster patterns to produce xrays, all aimed at a live human subject's mouth. If control was too slow, or hung, the focused electron beam would damage the thin tungsten target and let cooling water into the vacuum system, causing downtime for repair of weeks or more for a unique national shared speech research faciility.)
The best feature of wire wrap for prototyping was a connection change could be made in about a minute. Comments here about various techniques are interesting.

The same small group was involved in early design of at least one of the less costly VAXes, the 730 I think. IIRC both systems used AM29xx bit slice processors.

[xilman]

Originally Posted by kriesel

The same small group was involved in early design of at least one of the less costly VAXes, the 730 I think. IIRC both systems used AM29xx bit slice processors.

As a grad student, and then subsequently in my first paid employment, I worked with a team designing a 32-bit AM29xx machine which had comparable performance to an 11-750 at a third of the price. I wrote virtually all the microcode for the HLH Orion machine. This was roughly 1981-1983.

The Orion was prototyped with wire-wrap but ran at speed limited by the chips, not the interconnect, so was just as fast as the production circuit boards.