[Simh] The missing PDPs

[Johnny Billquist]

On 2019-08-13 07:55, Warren Young wrote:
> On Mon, Aug 12, 2019 at 8:17 PM Johnny Billquist <bqt at softjar.se 
> <mailto:bqt at softjar.se>> wrote:
> 
>     As far as I can
>     remember, the changes needed compared to a PDP-8 is not much. PC is at
>     address 0 instead of internal to the CPU, and interrupts then store the
>     PC at 1, and start executing at 2.
>     All the rest of the CPU works the same, as far as I remember. 
> 
> 
> There's a lot more to it than that. See the link in my prior reply in 
> this thread. For one thing, there's the matter of combined OPR instructions.

That is already an issue depending on which PDP-8 model you are 
emulating. Not all models allowed any random combinations (well, more 
properly didn't produce the expected results). Also, some combinations 
were undefined on some models, but produced usable stuff on other models.

> It is *possible* you could have PDP-5 code that will happen to run on 
> SIMH's PDP-8 simulator with little to no modification, but there is 
> probably also code that pokes at these known differences to determine 
> whether it's running on a -5 or an -8 and behaves differently if it 
> thinks it's on an -8, which would obviate the purpose of having a 
> "PDP-5" simulator.

Almost all PDP-5 code will run on a PDP-8, with the notable exception of 
any code which uses interrupts, or use address 0 for something.

If you have code that also tries to detect what it is running on, it 
will most likely not think it's running on a PDP-5 if it is running on a 
simh PDP-8, obviously. :-)

But the "interesting" case is, I guess, any PDP-5 code which do use 
interrupts. Any such code will not work in simh today, and for those 
programs, simh really will need a proper PDP-5 implementation, which is 
separate from the PDP-8. But cloning the PDP-8 is probably a very good 
starting point.

>     there might be some peripherals to work on, which any PDP-5 software
>     would be expecting...
> 
> 
> I think that's highly likely.
> 
> I suspect that unless you're running some bit of software that *does* 
> depend on some PDP-5 era device that the current SIMH PDP-8 simulator 
> doesn't bother to emulate, you're probably not providing much value in 
> having a PDP-5 simulator in the first place. That software might just 
> run atop the current -8 simulator, and then what value have you wrought?

Like I said. I think a PDP-5 should be rather simple to add, since it is 
so close to the PDP-8. But there is a crucial difference in how 
interrupts work, and how the first 3 locations in memory are utilized, 
which makes a PDP-5 different than a PDP-8. And that means there are 
definitely PDP-5 software that will not work on a PDP-8. As far as 
peripherals go, it needs to be looked at a bit. Console is the same, and 
I believe paper tape is too, which are the two most crucial bits. I 
can't remember what kind of mass storage might have been used on a 
PDP-5, or if any sort of more sophisticated operating system existed.

>     The PDP-12 poses a more interesting problem, as it is a dual CPU thingy
>     with shared memory and two totally different instruction sets, but some
>     interaction between them...
> 
> 
> And then imagine the eventual PiDP-12, complete with lab I/O 
> peripherals, scope driver...mmmm. :)

It would be fun, indeed.

   Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt at softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol