Possible to speed up I/O subsystem of 5150?

[Steve Thatcher]

the real ATA spec was not "approved" until 1990 or so - the hazards of 
being the first standard for such an interface. The ATA-1 and ATA-2 at 
least had a hardware line going to the drive to switch between 8 and 16 bit 
data paths to the drive. Any new drive would have to indicate compatibility 
with ATA-1 or ATA-2 to work with an older controller. It loosk like ATA-3 
or 4 did away with an 8 bit transfer capability. Obviously no one is making 
ISA IDE controllers anymore, but one could design one that worked perfectly 
well with the new drives. One does not have to assume that the data bus to 
the drive must match the data path to memory

At 04:43 PM 12/09/2004, you wrote:

>  I thought they were somehow "ATA compliant".  I
>assume you're talking about the interface between the drive and the
>controller, right?
>
>--
>
>The issue is what data bus width is assumed during DMA transfers.
>16 bits is 'normal'. Apparently, it was possible to do 8 bit DMA
>transfers on early drives. I've not looked at the interface for
>CF. Guess it's possible 8 bits is supported there as well.