[vcf-midatlantic] Original Macintosh Architecture Questions
John Heritage
john.heritage at gmail.com
Thu Jun 24 20:05:54 UTC 2021
Thanks David!
That diagram (Apple Mac Logic) explains perfectly why the 1990 Mac Classic
and the Atari ST were often 20-25% faster than the original Mac despite all
being 8 MHz 68000s.
The Original Mac gave 4 cycles of memory for CPU followed by 4 cycles of
memory for everything else.
The ST gave 2 cycles of memory for CPU followed by 2 cycles of memory for
everything else (glue/shifter/mmu).
The upgraded Macs basically gave 6 cycles of memory for the CPU then 2
cycles for everything else by doubling memory speed.
The 68000 would stall for some (many?) instructions with the 4/4 cycle but
is able to more or less run at full speed on the 6/2 (Mac) or 2/2 (ST)
cycling.
Interesting & Thanks!!
Last question on the Mac Audio --
Is it fair to say that if you wanted audio on the Mac, you basically wanted
to digitally sample a sound/take a sound sample and then have the CPU shape
it so it would output correctly? (i.e. costing some CPU cycles)
On Thu, Jun 24, 2021 at 11:21 AM David Riley <fraveydank at gmail.com> wrote:
> On Jun 23, 2021, at 4:24 PM, John Heritage via vcf-midatlantic <
> vcf-midatlantic at lists.vcfed.org> wrote:
> >
> > Hey folks --
> >
> > Two items I'm curious about on the classic (and very original) Apple
> Macs..
> >
> > 1. Mac (128kb ,512kb, etc) audio -- was there a custom or off the shelf
> > sound chip for the audio on the original Mac? and was it 1 channel / and
> > was it a custom chip by Apple? (how did it achieve speech synthesis in
> > 1984)
>
> It used some extra gates on the PALs in the timing and/or memory
> controllers to do a PWM which was then filtered down by an RC circuit to
> make a crude (but effective) DAC, similar to how delta-sigma DACs work. A
> similar circuit did the speed control for the floppy drive, which slowed
> down as it approached the outer tracks. Both took the PWM width data from
> unused bytes at the end of the video rows as the video scanning circuit
> went through horizontal refresh, so there weren't any real wasted cycles.
>
> > 2. The Macintosh Classic released in 1990 "was about 25% faster than the
> > original Macs". This is in line with the Atari ST often being quoted as
> > faster than a Mac when emulating a Mac via Magic Sac or similar. What
> > changed in the architecture to make the Mac Classic (or an ST in
> emulation)
> > faster than the original Mac? Was it faster memory that gave more
> > availability for CPU usage? were only certain instructions sped up or
> was
> > everything faster?
>
> It was actually improvements in the bus timing PALs that made the cycles a
> little tighter, allowing the CPU to run faster because it wasn't waiting
> for RAM as long. Jecel Assumpciao has/had some really good stuff on the
> memory timing generator (including the sound/floppy PWM) here:
> http://archive.retro.co.za/mirrors/homepage.ntlworld.com/kryten_droid/AppleMac/apple_mac_logic.htm
>
> In any case, it was still an 8 MHz 68000, but the improved memory timings
> did yield a performance improvement.
>
>
> - Dave
More information about the vcf-midatlantic
mailing list