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

[Andrew Back]

On 13/09/2023 00:31, Dave McGuire via rescue wrote:
> On 9/12/23 18:41, Joshua Boyd via rescue wrote:
>> If we are going to call the XT/370 an emulator, then isn't a Motorola
>> 68000 chip just also an emulator?  As I understand it, when you get to
>> microcoded chips, there is no actual hardware that runs the actual
>> binary instruction set sent to the processor. With either the regular
>> 68k or the 370 implemented in 68k microcode, the CPU reads an
>> instruction of the either 68k or 370 origin, then looks that up in ROM
>> for how to execute it.
>>
>> Now, is a gate level re-implementation of a 68k on an FPGA emulation? If
>> so, then I suppose that means anything on an FPGA is emulation. Is a
>> blackbox external "reimplementation" of a 68k on an FPGA or ASIC
>> emulation? The MiSTer people, to my understanding, state that it isn't
>> but I feel suspicious about that, and would probably err on calling that
>> emulation.  What if it was a blackbox re-implementation but could run
>> original microcode? I probably don't want to even debate that one.
> 
>     We are rehashing the age-old argument of "what is emulation".  All
> IBM S/390 processors are microcoded; the micromachine doesn't understand
> the S/390 instruction set until the microcode is loaded from the console
> system.  This is the case for nearly everything; most processors are
> microcoded.  You know all of this, having stated that you don't want to
> debate it. B-)

There is a nice IBM presentation, The What and Why of
System z Millicode.

https://www.cmg.org/wp-content/uploads/2016/08/the-what-and-why-of-system-z-millicode.pdf

It was interesting to learn that hardware virtualisation support 
(logical partitioning) is implemented in microcode, and seemingly this 
is based on IBM VM, the software product which started out under that 
name on System/370 (and before that as CP/CMS and CP-40 on System/360).

Andrew