Re: Apple II GS Question

[pubpc1@library.ucla.edu]

dempson@actrix.gen.nz (David Empson) wrote:
in response to my reply:
>
>> [slot speeds]
>>
>> On the Zip, you are given a choice.  You can set each slot to so that
>> the CPU does NOT slow down to 1MHz when accessing it.  Doing this
>> allows the Zip to maintain its fastest CPU speed (say, 15MHz) while
>> accessing the slots.  As you can imagine, setting a 5.25" floppy
>> drive's slot to accelerated speed makes it unusable, but setting a
>> hard drive controller's slot to fast makes the hard disk run faster.
>
>Sorry, Scott.  You are mistaken.

Oh, I'm sorry.

>
>There is no way that the ZIP can bypass the 1 MHz bandwidth limit on the
>slots, as it is controlled by the FPI (or CYA) and Mega II (both on the
>motherboard), which the ZIP must cooperate with.

I guess I was thinking of DMA cards.  Seeing that DMA bypasses the CPU,
should say the Apple II High Speed SCSI card be immune to this
limitation?  I would think that the Apple card would be pumping data
at up to 1MB/sec into main RAM while the Zip could continue to operate
at 15MHz, providing that everthing is cached at the right time.

>
>The slot speed setting for the ZIP determines whether the ZIP enables a
>delay when any I/O location in the slot is accessed.  (Probably also
>firmware locations - I haven't experimented, and the technical
>documentation is vague.)
>
>If the slot speed is set to "fast", the cycle in which the slot is
>accessed will be at 1 MHz, but subsequent cycles can resume fully
>accelerated operation (assuming no other delay is active, and the
>accelerator is enabled).
>
>If the slot speed is set to "slow", the cycle in which the slot is
>accessed will be at 1 MHz, and the ZIP is disabled for the next 50
>milliseconds, causing subsequent cycles to run at 1 MHz, 2.6 MHz or 2.8
>MHz depending on the memory area being accessed.  If another access to
>the slot occurs during this time, the delay is retriggered.

Here's a question: what does disabling "Control Panel Follow" do?
The Zip documentation suggests that it ignores the motherboard's
request to slow down to 1MHz.  Ever since I started using an FDHD
floppy, I have this setting disabled and everything moves a little
faster.

>
>
>The reason that there is a problem with 5.25" drives (and some other
>hardware) is that software which is accessing the I/O locations is
>performing timing-critical operations, relying on the speed at which the
>CPU executes the instructons.  The accelerator would throw this
>completely out of kilter, so it is necessary to temporarily disable the
>accelerator whenever the appropriate slots are accessed.
>
>Some I/O devices are not timing-critical, so there is no need to slow
>down the software (e.g. most parallel cards should work fine with the
>slot set to "fast").

That makes sense.

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