Hand-wired core for ROM?

[Eric Smith]

On Sat, Apr 11, 2015 at 11:13 AM, Marvin Johnston <marvin at west.net> wrote:
> The PDP16 had a constants card circa 1972 or so. The constants card was
> basically cores with hand threaded wire done by the customer, and acted like
> a ROM. I don't remember the capacity, but IIRC it was about 16 bytes.

That was the MR16-D "constants memory", M7325. It stored 24 words of
16 bits each, so the overall capacity was 48 bytes. The module sold
for $150 in 1972.

Unlike the other PDP-n machines, the PDP-16 was not a computer with a
standard instruction set, but rather was essentially a set of Register
Transfer Modules (RTM) primarily intended for building hardwired
systems using asynchronous control logic, though there were also
16-bit data path and memory modules, including larger ROM modules that
could be used as a microcode control store. The asynchronous control
flow worked in a similar manner to many early PDP-n machines, where at
any given moment usually only one functional was active, and when it
was done, as it became inactive it sent a pulse to the next functional
unit. Branches in flow were done by having a unit decide which of
several units to send the pulse to. In the PDP-16 RTMs there was also
provision for subroutines.

However, DEC did offer a preconfigured PDP-16/M computer, which was
built from the PDP-16 RTM modules, but used them to implement a more
conventional 16-bit processor which did have a standard instruction
set (though not matching that of any other DEC machine). A normal
PDP-16/M used the PDP-8/M chassis and power supply, though it could
also be purchased as only the backplane and modules.