[rescue] [OT] S: IBM /370 or /390 card

[Mike Katz]

I used to work for Software Development Systems.  We made 68000 C 
compilers and emulators.  We were doing cycle accurate simulation of the 
68000 and other 68K CPUS up to but not including the 68020.  The 
pipeline made things more difficult to do cycle accurate simulations.

Yes this is different from writing a micro-code emulator.

On 9/12/2023 4:45 PM, Joshua Boyd via rescue wrote:
> On 9/12/23 16:37, Dave McGuire via rescue wrote:
>> On 9/12/23 16:25, Mouse via rescue wrote:
>>>>> Wasn't the XT/370 primarily made with 68000s with custom microcode?
>>>>> [...] I'd love to see someone work out how to redo that on a newly
>>>>> made board.
>>>> Since the 68K's control store is in mask ROM on the chip, I think
>>>> that would be a challenge.
>>>
>>> I can't imagine that there would be any difficulty beyond possible lack
>>> of documentation in reimplementing a 68k with writable microcode in a
>>> modern FPGA.
>>
>>   How I interpreted Josh's question was the modification of a 68K's 
>> microcode.
>
>
> I was originally thinking modify the microcode in the chip, but making 
> a cycle accurate microcode compatible 68K in an FPGA to run altered 
> microcode would also be cool.  Extra cool if the FPGA's image was 
> derived from a scan of the 68k silicon.  Square the cool if that was 
> then run through on one of Google's Open MPW shuttles.
>
> I haven't dug into it too deeply, but I think that when people talk 
> about emulating 68000s as part of, say, the MiSTer project, they are 
> using a fairly high level emulation of 68000s, not something that is 
> executing original microcode. I guess it would be interesting to know 
> what is really happening there.  Also how far down the accuracy hole 
> the MAME 68k core goes.
>
>
>
> _______________________________________________
> rescue list - http://sunhelp.org/mailman/listinfo/rescue_sunhelp.org