Alpha CPUs, was Re: [rescue] i860 Success

[Francisco Javier Mesa-Martinez]

On Mon, 3 May 2004, Phil Stracchino wrote:

>
> On Mon, May 03, 2004 at 08:09:44PM -0400, Nathan Raymond wrote:
> > On Mon, 3 May 2004, Phil Stracchino wrote:
> > > (However, I don't personally know whether the 68000 supported virtual
> > > memory.)
> >
> > http://www.informationheadquarters.com/Apple_Macintosh/Motorola_68000.shtml
> >
> > "The 68000 was also unable to correctly return from an exception on a
> > failing memory access, a crucial feature to enable true virtual memory.
> ....
> > "In the next major revision, the 68010, most of these problems were
> > fixed."
>
> That coincides with my recollection that the first CPU Apple shipped
> with virtual-memory capability was the 68020.

I believe it was because Moto also offered the 68020 as a full chipset,
including math coprocessor and MMU. Before that most people had to use the
68K with their own MMUs (like SUN). I believe that even with the 020 SUN
used their own MMU (but I may be not correct there). The 020 was full
32bit adress, where the 68000 was 24bit real adress, but it used
internally 32bits I believe, I dunno if this arrangement made the 68000
hard to deal with. But at least it allowed any program to use 16MB which
in the early 80s was a pretty big deal. I suppose that is why the original
macOS came in 24 and 32bit modes.

Motorola seemed to have a much cleaner architecture, as they got the 68000
out of the door with just 68000 transistors (and that is pretty neat for a
system with 32bit registers), at least a lot of the alternative
architectures from the 80s seemed to prefer Moto over intel. Atari, Amiga,
Mac, SGI, SUN, NCR Tower, HP 9K et al all were based around Motorola CPUs.
I guess the RISC transfer hurt them more than it did to intel :(. But I
guess maybe Motorola had more to do with the adoption of Unix in the
desktop during the mid 80s than intel. Although I dunno the actual figures
for Xenix, coherent or even sequent which were x86 based nixes.