Self modifying code, lambda calculus - Re: ENIAC

[mark at markesystems.com]

> Early 3rd generation machines had special instructions to finagle their
> way around self-modifying code:

And some didn't:  The HP 2100 and the PDP-8 (and I think the Honeywell
x16s), instead of a stack, would store the return address of subroutine
calls in the first word of the subroutine; obviously, this made recursive
subroutines impossible without handling a stack manually.

> Few CDCers of the time even knew that the STAR-100 existed.  I remember

Reading up on the early history of CDC, I stumbled across the "Little
Character" - Seymour Cray's six-bit proof of concept for his first computer
designs at CDC.  Apparently, one was actually built, because it's now at the
CHM.  However, I haven't been able to find any significant information about
this on the Web (my Google-fu must be failing me).

Does anyone know where I could find some documentation about this machine?
Performance specs (memory size, speed, etc.) would be nice, but I'd really
like a detailed architecture and instruction set description - you know, in
case someone wanted to make an emulator or something...
~~
Mark Moulding