Dartman wrote:And yes I belive the IDE chain will only run as fast as the slowest device connected.
That used to be the case for older IDE controllers, but modern controllers typically support independent device timing, so it isn't true anymore. For example, you can mix a UDMA-5 master with a UDMA-3 slave, and each will run at its own rate. Of course, it still takes longer for the UDMA-3 device in this example to transfer data, so when both devices are transferring data at once (actually taking turns using the bus) the overall rate will be slower than if both used UDMA-5, but it will be faster than than if both used UDMA-3. The master transfers at UDMA-5 rates while the slave transfers at UDMA-3 rates, making for an overall average somewhere in between. And if the UDMA-3 device isn't actively transferring data, the UDMA-5 device will have the bus to itself and transfer at the full UDMA-5 rate.
Contrast this to to a hypothetical older controller that doesn't support independent device timing. It would force the UDMA-5 master down to UDMA-3 rates even when the slave wasn't transferring data - the situation to which Dartman referred. (Note - a real controller that supports UDMA is new enough that you can be assured it does support independent device timing, which is why I called this a "hypothetical controller".)
cfitz