Bluh. And that said, obviously, any compatibilities/incompatibilities with IBMs would have to be traced back to whatever particular (on-disk) controller wouldn't play nice, and the particular models or model line using that design. If you compare a DFHS or something with a GXP (brought up again only because I don't know any other IBM models, heh), you may as well be comparing Quantum to Seagate - they're just utterly different, even though they're both IBMs.
Here's a thought on the issue - I picked up a certain Seagate from the era when Pentiums and x86 boards as a whole were transitioning from 5V to 3.3V signaling. I have no idea what the IDE/ATA/UDMA spec is supposed to be, but apparently, at some point, a transition in signaling voltage was made. There was a known problem with the (pre-UATA) Seagate, defined as incompatibilty when run on a specific newer (UATA) Intel chipset (whichever ones used the 'PIIX4'? Ugh, been a while), because it couldn't deal properly with the reduction in voltages. Somehow, I bet that model won't work on any other UATA controller made since, but Seagate was probably holding out hope that future chipsets would detect and go back to 5V swing for older drives... and maybe they do. :-? Modern drives are designed to tolerate a fairly wide range of swing, so they can fall back or forward as appropriate.
So, here's a thought. The AGA Amigas were making some transitions of their own in that department, weren't they? Maybe? And if not, then maybe IBM had the opposite problem on one of their drives, producing a UATA-era model that wouldn't tolerate things when placed on a 5V bus?
Of course, I'm probably totally wrong, and it's probably just mysterious IDE retardation of the same sort that kept early Conners from sharing a channel with other manufacturers' drives.
