LinuxSA Mailing list archives

  From: Mike Andrew <mikero@norfolk.nf>
  To  : linuxsa@linuxsa.org.au
  Date: Thu, 29 Jun 2000 12:08:56 +1130

Re: More space required

From: David Newall <davidn@rebel.net.au>

Hmmm, hope I'm not dragging this mailgroup off topic, if so, say so.

> The PC was by no means an order of magnitude better than Apple's low-end
> machines, and compared to Apple's newer machine, Lisa, I think the
> convincing argument is that PC was inferior.

Nicely put. Orders of magnitude? No. Not at all.

> I'm unconvinced that "the design of the PC at the time was excellent."

Oh yes sir, it was. For that matter so were Apples. For the technology
available (TTL SSI logic) all did a fine job. The overall design of all
these machines were excellent, with accompanying design flaws. Yes, you can
have a dichotomy.

There are a couple of corrections I need to make to what you said.

> system clock ran at 1MHz.  The Apple ][ had two interrupts: One maskable
> (ie the hardware could choose to ignore it) and one non-maskable (it could
> not be ignored.)

All processors have this nmi versus irq pin. Nothing new there. In the case
of other cpu's the single maskable interrupt was an electrically shared
signal amongst them all (active low). In the case of intel it required
active high, non shared, interrupt controller chips, forming a funnel (that
you mention). This was fundamentally wrong. The logical conclusion is, 1024
devices would require 1024/(8-1) controller chips. A design that in fact did
occur for serial multiplexers (300 and 2400 baud). Other architectures,
Motorola with it's four cpu state pins, and Zilog with it's int acknowledge
forced the responsibility back on the device to provide a vectored interrupt
address. It doesn't matter that 'a' cpu had two or more interrupt pins, that
design path was flawed to its very core and remains with us today (as you
say) with resource conflicts.


> Apple, Apple Apple

Apple rode on the back of a far superior chip design, the MC680x and
MC680x0. Motorola shot themselves in the foot by making the 68000 binary
incompatible with the 6809. Apart from their clearly superior OS design (gui
et al), they had it made with a 'better' cpu. This did not translate to a
better overall architecture (to the PC) because of Apple's ferocious
protection of 'their' design. Apple's greed with their pricing structure was
not just their undoing, they brought down Motorola as well who finally gave
up the race 486 vs 040. Most of the Motorola designers you would talk to
today would spit in the eye of Apple anything. They poured billions into
their 680x0 series to the benefit of Apple, who did not reciprocate (or
understand) that Motorola needed volume to cover the cost. Not proprietary,
restricted architecture. It is the unitary greed of Apple corp that has left
us with Wintel today. This has become so pathetic that (as I hark back on),
all we can discuss now is the relative cooling properties of a flawed cpu
and gape with wonder and awe when Intel provide MMX bungleware.

Final comment about MC is that these cpus have NO input or output
addressing. Everything is linear memory. Today, we are stuck with a pc bus
that confuses itself between 10 bit i/o space, 16 bit i/o space, slow vs
fast i/o and uneccessary instructions to effect same. The C language pays
the penalty for this. In MC / Unix architectures we just use (memory)
pointers to everything. In x86 we need special non-c inportb outportb
instructions. The original K&R which got it right everywhere else, does not
mention inp() outp(), not imagining anyone could be stupid enough to design
such an animal I quess.

> The Apple /// system clock ran at 2MHz, and, like the IBM PC,
> included a standard real time clock.

The pc XT had no real time clock. It was introduced on the revamped AT
series. And, interrupts, remember them? changed for device assignments as a
result. Interrupt resources were designed wrong from the ground up,
necessitating even this change at such an early date.


http://users.nf/linux/ +mirrors
StepByStep submissions: mikero@norfolk.nf





-- 
LinuxSA WWW: http://www.linuxsa.org.au/  IRC: #linuxsa on irc.linux.org.au
To unsubscribe from the LinuxSA list:
  mail linuxsa-request@linuxsa.org.au with "unsubscribe" as the subject