My interest in classic computers is primarily in the architecture (i.e. -
programmer's model), not the actual metal, silicon, and plastic, although I
cannot deny a certain satisfaction in using at least parts of an original
machine to implement that architecture.
The PDP-11's I'm playing with at present are machines that are newly built
from scrap parts. I don't care whether those parts are DEC parts or not. I
am currently collecting parts to build a PDP-11 that will consist of:
- M7554-SD CPU board from a DECserver 550
- gutted SparcStation 1 (nice compact box and power supply)
- SCSI Qbus board from a microVAX (off eBay)
- SCSI hard drive (from an old PC)
- VT420 off eBay
- 100 pin .125 pitch backplane connectors (non-DEC) from Digikey.
It will run 2.11BSD. I consider this as legitimate a PDP-11 as any 19" rack
with pink and purple paint and blinkenlights.
Inconsistently, I have reservations about the legitimacy of my notebook
computer running simh and 2.11BSD, although so far I've done all my PDP-11
software development on this platform. I guess it's tainted by its too-close
contact with Windows. I'll continue to use it, but I won't be satisfied
until any software I develop is running on a "real" pdp-11.
An idea that I've been thinking about is the re-implementation of a classic
architecture (for me, probably a PDP-11) using entirely modern hardware. One
way to go about this would be to use a Xilinx FPGA. I know that this has
already been done for the PDP-8. Another approach would be to view a Wintel
PC as a microcoded platform, and reimplement the PDP-11 on that hardware.
The microcode would be written in tightly crafted Pentium code. The machine
would bootstrap itself by loading the microcode from a special partition on
the hard drive. Since the hardware has memory to burn relative to the
target, instruction decoding would be via huge jump tables. Don't bother
decoding anything, just use the entire 16 bit opcode as an index into a
table. The execution routines for each instruction would be generated by
macros or a C program, one routine for each of the 65536 possible opcodes.
The Pentium memory management hardware could be used to emulate the PDP-11
MMU. Accesses to the I/O page would be trapped by the hardware to routines
that mapped the PDP-11 registers onto the real available devices. Although I
could probably do this, I'll probably never get the chance -- it would be a
pretty ambitious hobby project. It could be done for any architecture. I
would consider such a reimplementation to be a legitimate instance of a
classic computer.
I don't see that there is a need to change the charter. Nobody pays
attention to the charter anyhow. ;-)
--
Jonathan Engdahl???????????????? Rockwell Automation
Principal Research Engineer????? 24800 Tungsten Road
Advanced Technology????????????? Euclid, OH 44117, USA
Euclid Labs????????????????????? engdahl_at_cle.ab.com 216-266-6409
http://users.safeaccess.com/engdahl/pdp-11.htm
Received on Fri Sep 14 2001 - 07:18:58 BST