Ruud@baltissen.org wrote:
> The IEEE drives have two CPU's onboard, the 1541 only one. The one 
> in the the 1541 does in fact the same job but now on its own. The 
> DOS of the 2031 is derived from the 4040 most probably. The DOS of 
> the 1541 is derived from the 2031 on its turn. That's why you still 
> find references to two drives.
> 
> The two main reasons not paying much attention to, let's call it 
> 8250IDE, is that 1) I have CBM-HD that forfills all my needs here 
> and 2) 1541IDE addresses a larger user group IMO. 
> 
> But because of this 90x0 discussion I paid more attention to it. And 
> I started wondering: why do the dual IEEE drives need two CPU's? Do 
> they REALLY need it? The 1541 prooves it can be done with one CPU. 

The 154x series has a serial bus, which is a slower version of the IEEE
488 parallel bus (GPIB). In terms of protocol, they're very close (ATN,
EOI, etc.), however the Commodore serial bus is very different in terms
of signaling those logical conditions.
  http://en.wikipedia.org/wiki/IEEE-488

The 1540 was an attempt to cost-reduce the PET drives while not changing
the basic messages. They were able to get rid of one CPU and build a
really cheap plug/cable standard. However, it would not be possible to
handle a second drive with this design. The additional GCR decoding
alone would exceed the CPU speed, leaving no room to do the drive
control. Remember that these drives had to support drive-to-drive copy
with no intervention from the PET.
http://www.commodore.ca/manuals/commodore_1541_4040_8050_8250_comparison.htm

If you look at a protection scheme like V-MAX, you can see how close
they were to the limit with standard GCR ona 154x. The standard sync
length of 40 bits was plenty long, but it took self-modifying code to
decrease this to the minimum of 12 bits (read-only).
  http://markus.brenner.de/mnib/vmaxtech.txt

So using this as an estimator, one drive with one CPU could barely hit a
3x reduction in sync length. It's extremely unlikely that it could
handle two drives (control + GCR) that were operating at the same time
(e.g., drive-to-drive copy).

-- 
Nate

       Message was sent through the cbm-hackers mailing list