Alex Freed <alexf@mirrow.com> wrote:
"David Empson" <dempson@actrix.gen.nz> wrote in message
1hsc1jh.1hqbl6aapwbr5N%dempson@actrix.gen.nz">news:1hsc1jh.1hqbl6aapwbr5N%dempson@actrix.gen.nz...
I wonder if the Unidisk could be used as a co-processor,
doing non-disk access computations?
In theory, yes. It would be a rather pointless exercise, unless you were
transferring a relatively small amount of data, doing a long and complex
calculation (with the Apple II doing something else at the same time),
then transferring back the result (also a relatively small amount of
data).
The overhead of the data transfer wipes out any gain in having access to
a second processor, except when the computation time outweighs the
transfer time.
One application that comes to mind is on-the-fly encryption of
data going to/from the disk. The usefullness of this in A2 context
is a different story.
Even then, you don't gain much by doing it in the UniDisk 3.5, unless it
really is running at 2 MHz - in that case it would be able to encrypt
the data twice as fast as an unaccelerated Apple II.
One could argue that doing the encryption in the UniDisk is harder to
crack, because the code doing the encryption has to be downloaded into
the drive, but that code has to exist on the Apple II in the first place
to get it onto the drive, so reverse engineering would still reveal the
algorithm.