PiStorm

[stephen_usher]

On the QL you wouldn't be able to fit a PiStorm into the case... and anyway it's not an 8 bit multiplexed form factor, so it would have to go onto an external board.

Hardware is the smaller part of a potential QL PiStorm project. It could be redesigned to fit the QL case.
I'd expect 90% of labour and problems on the software side.

The Pi would always be too tall to fit inside the BBQL except within the expansion slot.

As for the software side, if it merely emulated a stock 68000 with no emulated memory, just accessing the system bus then the software side would likely be minimal as the timing side of this and bus arbitration have been worked out on the Atari ST port. Similarly, adding RAM and even a copy of the ROM shouldn't be massively difficult. It's when you go further and try to add other peripherals is where you get into difficulty.

Forget the Amiga PiStorm code, that's niche as the Amiga ignores most of the important parts of the processor, moving all that complexity into the custom chips, making the CPU emulation simpler as it doesn't have to implement most of the timing critical parts.

[pjw]

My point is that MC68 CPUs arent being produced any more and as things look
now, will forever be stuck at 60 to 80 MHz and be as rare as hen's teeth
and cost too much.

But say some chip manufacturer came up with a brand new, low-powered, 100%
compatible MC68020 clone running at 300MHz for $20 apiece. Wouldnt anyone
here like to design a new "QL" around such a processor? A new "Q3X" or
"Q88" or whatever? I think that would be rather nice! Id want one!

CPUs run on microcode, so who is to say that if it runs on one kind of
microcode it is genuine, and on another it is not? I wouldnt mind if the
CPU were emulated on a much faster chip, such as ARM, provided it were a
drop-in replacement for a corresponding, but much slower, genuine MC68K
CPU.

I dont know if that is even possible. Thered have to be some jiggery-pokery
going on to replicate the signals on the pins, so some additional circuitry
would no doubt be required.

Has anyone considered ColdFire? (I know Nasta has..) Some variants can
reportedly be overclocked to 300MHz. I understand the instruction set is
not 100% compatible, but as far as SMSQ/E is concerned this could possibly
be worked around. As for old, sourceless software it could possibly be
managed by trapping illegal instructions. And some software wouldnt work at
all, but thats the price of progress..

[Peter]

The Pi would always be too tall to fit inside the BBQL except within the expansion slot.

There are several options which fit, the nicest being a board for the CM4.

[Peter]

Has anyone considered ColdFire? (I know Nasta has..)

That the very first chip in the Coldfire line but no actual Coldfire CPU. It is a 68EC40 CPU with multiplexed bus. Nice chip, but not if you want to be much faster than a 68060.

Some variants can reportedly be overclocked to 300MHz.

You lose a lot of that by the emulation of the missing instructions. And you have misimplemented instructions that don't trap out. A can of worms for QL code. Also look at the Atari related Coldfire hardware and check their benchmarks running real 68K code. Not overwhelmingly better than a good 68060 design.

[stephen_usher]

The Pi would always be too tall to fit inside the BBQL except within the expansion slot.

There are several options which fit, the nicest being a board for the CM4.

Remember that it would have to have a custom PCB which adds multiplexing onto an 8 bit bus.

[Peter]

Remember that it would have to have a custom PCB which adds multiplexing onto an 8 bit bus.

As mentioned earlier, I meant a redesigned QL PiStorm board. Which would of course have an 8 bit bus. Not something using the Amiga PCB. But a smaller form factor isn't needed for the initial prototype. That could be a GoldCard with CPU socket adaptor to the Amiga PCB or whatever. Before the software job has been done, the final form factor is not important. Let's see if someone is motivated enough for all the work and debugging.

It is much easier to just let the 68008 feed QL keyrows into a Raspi and use sQLux. Function-wise, a QL PiStorm is not likely to offer more. The PiStorm motivation I see here on the list seems rather academic interest.

[Derek_Stewart]

Hi,

Another solution would be to connect the Raspberry Pi to the QL data bus, so that the Raspberry Pi could be a co-processor to the QL.

The main problem, is writing the device drivers, on the QL and maybe bare metal code on the Pi.

Sort of like the BBC Tube interfacd for the QL, which maybe as realistic as a PiStorm on the QL connected to the CPU socket

[Dave]

The Pi would always be too tall to fit inside the BBQL except within the expansion slot.

I have a customer Pi-based board using a CM4 that is 9mm tall at its highest point. Everything in PiStorm can fit on a baseboard with a CM3. A CM4 (if PiStorm supports Pi4 yet) would be 0.7mm slimmer.

So a PiStorm could physically fit inside the QL case with the QL PCB. If a SPEM memory can do it, a PiStorm can.

That said, a Pi400-like board inside the QL case seems a far more likely occurence - especially as there are already actively developed and supported emulators that can run on it today.