Now, the question is how to define 'Non-PC
compatible'. After all,
the PC itself isn't that hard defined. If you look on what machines
I would argue the PC _was_ defined by the IBM Techref.
DOS was running, including highly complex applications
like Word,
Autocad or even Windows, you'll find that the common denominator
often just the use of a x86 CPU is.
There's a lot more to a machine than the CPU.
Go back to the time of the IBM PC. There were several other 8088 boxes
around that were not compatible. Sure they might have run MS-DOS or CP/M
86. But you couldn't take a bootable disk from one machine and boot it on
another. In many cases applications programs from one machine wouldn't
run on another,
I seem to rememebr at least 3 levels of compatibiliy :
OS compatible -- the MS-DOS calls were the same, any program that used
those and those alone would run on both machines
BIOS compatible -- the BIOS calls were the same. You could use those (and
MS-DOS calls) and have no compatibility problems
Hardware compatible. The memory map was much the same, I/O devices had
the same registers at the same port addresses, the video system was the
same, and so on. You could 'hit the bare metal' with no compatability
problems.
It is the last level that we normally consider when we talk about 'PC
compatible' I think,
Thus every Pentium based computer would be PC-Compatible by definition.
By that defitinion _any_ computer is PC-Compatible. My HP71 could run an
interpreter written in BASIC that treats a couple of 9114 floppies as the
main memory of an emulated PC, and boots MS-DOS from a third one.
Accesss to video RAM could be sent to the HPIL video interface, and so on.
The fact that it would take years to even get to an MS-DOS prompt is
irrelevant :-)
But OK, I'll rephrase the question. Are there Pentium-based machines that
can't run the standard Windows we all know and hate, or a slightly
modified version of said software (and for reasons other than the trivial
ones of insufficient memory or disk space etc). That is, I am thinking of
machines with very different I/O devices, etc.
It appears such machines do exist. And I would argue that once they reach
sufficient age they should be considered as 'classics'.
-tony