PDP-8: (and others) anyone tried NVRAM simulation of hard drives?

From: Ethan Dicks <erd_6502_at_yahoo.com>
Date: Wed Feb 5 14:39:00 2003

--- ben franchuk <bfranchuk_at_jetnet.ab.ca> wrote:
> Ethan Dicks wrote:
> > My expertise is more in microprocessors; I was contemplating
> > throwing an MC68000 on the other end of the bus since I have extensive
> > stocks (dozens of tubes), a large assortment of hardware debugging
> > tools (code trace analyzers, a Flike 9010A, etc.) and lots of expertise
>
> So what is wrong with a 8 bit cpu since a 68000 is overkill.

Nothing, but as I said, I have the skills and the tools. OTOH, hooking
a 16-bit data bus to a 12-bit bus is a little easier than hooking an
8-bit bus to a 12-bit bus (having worked on a VAXBI product with a
68010 - that was a nightmare - 16 bits to 32 bits... can rant about
that one for a while if you have the time ;-)

> BGmicro has surplus 512k x 8 flash $1.00 each if need cheap
> memory but they have real HD's for $21.50 each. Now my question
> is" are you designing a new interface or emulating a old one".
> Ben.

Perhaps a little bit of both. If I were going to emulate a DF-32,
that could be done in battery-backed SRAM - 32K words per unit,
4 units per CPU. Trivial. The original logic was implemented in
R-series logic. It wouldn't be hard to do it all in TTL, just a
bunch of work. The actual storage technology isn't the issue; the
presentation to the PDP-8 is.

The advantage of emulating old hardware is compatibility with existing
drivers. The advantage of creating new hardware is freedom from ancient
and typically arbitrary constraints. The disadvantage is a) writing
the drivers (a one-time problem) and b) getting the new drivers onto
an old system (a one-time-per-CPU problem).

The SBC-6120 solves these problems by a) having had OS/8 drivers already
written and b) having enough smarts in the monitor to stuff a disk image
down the console serial line. Doing this for a real PDP-8/i would not
be as painless. Having a "real" processor on the peripheral card might
help mitigate these problems (i.e., partition, format and populate the
"disk" via the embedded CPU, then all you need is a bootstrap from the
-8/i side and off you go).

That's another issue... the -8/i has no ROM. Some of the older devices
had extremely short bootstraps (the RK8E is the shortest I know of - one
or two words). I would hate to have to toggle in a 100-instruction boot
everytime I corrupted memory. That points to emulation of existing
devices being a good thing.

> Personaly I have hard time dealing with a CPU on a i/o card
> that is more powerfull than the PDP.

Why? We used to sell them all the time. When COMBOARDs sold in bunches
like grapes (c. 1982-1984), they frequently went into PDP-11s running
RSTS because we could hang a line printer (LA-180 or larger) directly
off our card and have print jobs coming back from the IBM come in the
serial port and out the parallel port, never pestering the RSTS print
queues. Folks *loved* that. The CPU on the COMBOARD is an 8MHz 68000 -
_roughly_ equivalent to the CPU in the VAX-11/750 (~.6 VUPs) and quite
a bit faster than most PDP-11s of its day (not sure about the 11/70)

There is a period SCSI card for OMNIBUS PDP-8s - it has a 6809
processor on it to relieve the PDP-8 of the nitty-gritty details
of how SCSI works. Same idea, slightly newer peripheral hardware.

-ethan
Received on Wed Feb 05 2003 - 14:39:00 GMT