On Thu, 24 Sep 1998 Paul Sprague <[log in to unmask]> wrote:
><snip>I have my LPT 1 port using IRQ 7 and IO range 0378-037F.
>I also have a standard ISA expansion card from SIIG
>for LPT 2 which I use for my parallel port Zip drive
>using no IRQ and an IO range of 0278-027A.
>Everything works great. How is it possible for
>LPT 2 to function properly without an IRQ assigned to it?

There are two ways for software to service I/O operations:
interrupts and polling.
Each has advantages and diasadvantages.

Interrupts prompt the processor for service when the external
device needs it, but have a high overhead in saving and restoring
the register status of the processor.

Interrupt control is good for activities that need rapid attention
to continue running, (such as a data transfer with a disk or tape
drive or serial UART), but bad for very small and frequent activities
(such as sending characters to a parallel port) because of the high
processor overhead.

Polling is good for frequent, small transfers (because of the low
overhead) if the device doesn't mind waiting until the software
gets around to giving it some attention. If the device is busy and
can't accept a transfer, it wastes processor cycles in testing.

So for printing, polling is excellent because the processor overhead
is low and the printer can wait indefinitely for the next character.
The polling can be run as a background task so it gets the leftover
processor cycles which would be wasted anyway, it stuffs the printer
buffer until the printer goes busy, then idles polling the printer
until another interrupt (such as the time slice expiring) takes
away control. No printer interrupt is needed.

Parallel EPP and ECP modes are designed to be useful for more
sophisticated devices which might need urgent attention, for
example a CD-R burner. Although these devices have a large
internal data buffer, they can't wait indefinitely.
Without an interrupt to get attention, there would be a risk
of a data overrun (and making another shiny beer mat :-()

Your Zip drive is designed to work with a simple parallel port
for maximum portability, but if you can find it an interrupt and
EPP (or ECP?) and reconfigure by running the port accelerator
program it should run _much_ faster (possibly 2-3 times).


Len Warner <[log in to unmask]> http://easyweb.easynet.co.uk/~len/ ICQ:10120933

                                  -----
       **Need help with PCBUILD mailing list? Send an Email to:**
        Bob Wright<[log in to unmask]> or Drew Dunn<[log in to unmask]>