Re: Q: Timing: How many clock-cycles to print one char on screen?

[dempson@actrix.gen.nz (David Empson)]

Holger Picker <pickerh@uni-muenster.de> wrote:

> Thanks a lot for your help. That was useful information indeed. If you
> don't mind I'd like to add a question concerning the clock-stretching
> mentioned by David Empson. If I understood it right, then the
> videocontroller needs 65 "clock-cycles" compared to the processor to draw
> a line on the monitor.

Correct.

With the stretched cycle, an NTSC IIe has a horizontal scan frequency of
almost exactly 15700 Hz, which is close to the NTSC standard of 15734
Hz.  It takes 63.695 microseconds to output one line to the video
display.

The PAL version of the IIe uses a 14.25045 MHz crystal (the PAL IIc
apparently uses a 14.25 MHz hybrid circuit).  Allowing for the stretched
cycle, this gives a 63.998 microsecond line output time for the IIe and
64.000 microseconds for the IIc (it is supposed to be 64.000).

> Then 262.5 lines per frame * 65 clock-cycles = 17062.5 clock-cycles per
> frame * 60 frames per second (NTSC) = 1023750 clock-cycles per second ~=
> about 1.0227 MHz with a master clock frequency of 14.31818 MHz.

The Apple II generates non-interlaced video with exactly 262 lines for
NTSC (including the vertical retrace and border areas), so the correct
calculation is 262 * 65 = 17030 CPU cycles per frame, which is a period
of 16688 microseconds.  There are actually 59.92 frames displayed per
second (the ideal value is 59.94).

(All of the above figures are verified by "Understanding the IIe".)

> > There is a complication: to comply with NTSC video timing, the Apple ][
> > stretches every 65th CPU clock cycle by two cycles of the 14 MHz clock.
>
> Ah, I didn't know that (never heard of it)! What does it mean to the
> processor?

Not much - the CPU itself couldn't care less what the clock frequency
is, as long as isn't too high or too low.  The 6502 is driven by an
external TTL two-phase clock.

It may have implications in precise speed calculations, but it only
affects the execution speed by 0.2%.  I'd be more worried about hardware
implications, but there isn't anything critical enough for this to be a
serious problem.

> Would it  mean that after 65 clock-cycles the processor does one/two
> (complete) wait-cycle?

The processor is completely unaware of it - it just has one cycle out of
every 65 in which the clock which drives the processor has a longer
period for that cycle.

> Or is the difference too small to be noticeable (hearable? What about
> processor loops that access  $c030? Could you hear a "background-humming"?

If you were trying to generate a precise tone, you would need to allow
for this when determining the number of CPU cycles between accesses to
$C030, i.e. work on the assumption that the CPU is executing uniform
cycles at 1.0205 MHz instead of 1.0227 MHz.

I doubt that the "jitter" introduced by the regularly spaced stretched
cycles would be audible.

> I've always thought that the 6502 in an  AppleII runs at constant speed.
> Now it seems it was only true for PAL-Apples.

Nope - I was wrong on that count.  All Apple II models have this
irregularity in the nominal 1 MHz clock.

> And, if I may ask, what happens on an AppleIIc+(4Mhz) or an accelerated
> AppleIIe if you access the disk-drive (especially writes)?

The accelerators are all designed to deal with I/O accesses to speed
critical devices by slowing down and running in sync with the
motherboard during I/O operations.  In the case of the IIc+ (and the Zip
Chip, which is the basis of the IIc+ accelerator) there is a 50 ms "slow
down" timer triggered on each access to a slot I/O location for a slot
which is configured to operate in slow mode (such as the 5.25" disk
controller).

For some devices the hardware accesses are not timing critical, so the
CPU can be left running in accelerated mode, with only the actual I/O
cycles operating in sync with the motherboard (at 1 MHz).

> P.P.S.: Hmm, I think I had better put up another (earlier) post also in
> comp.sys.apple first so to show you what I'm driving at (emulator
> programming). Sorry, I'll do that at once.

I think I saw you post something on that in comp.emulators.apple2.

> As I live in Germany I probably will have no access to these books at all.

I live in New Zealand, and I ordered my copy of Understanding the IIe
from Byte Works.  They posted it to me, but I don't remember whether it
was airmail (quite expensive) or surface - the latter is more likely.
(This book is a softback, about 8.5 by 11 by 1 inch, or 215 x 280 x 25
mm.)

-- 
David Empson
dempson@actrix.gen.nz
Snail mail: P O Box 27-103, Wellington, New Zealand