I've run a few tests with 4 Hard discs, connected all to one FastTrak 100 TX2. Config:

(1) IDE1 Master: WD 1200 JB
(2) IDE2 Master: WD 1200 JB
(3) IDE1 Slave: WD 800 BB
(4) IDE2 Slave: WD 800 BB

(1) and (2) are standalone drives, while (3)+(4) form one Raid-0 Array.

When copying from (1) or (2) to (3)+(4), I get easily 30 MB/s, althout in theory the 2 drives, which share one channel (which are (1) and (3) as well as (2) and (4)), should slow down each other.

Is it possible that these slow down affects which one can often see with such configurations are only caused by bad chipsets and bad chipset drivers, and that IDE itself is not the problem?

That FastTrak controller seems to be able to access 2 drives sharing one channel pretty well.

It's astonishing how fast SupRip and DVD2AVI run with that configuration

Copying from one drive to another on the same channel is one operation, not 2, so you should get the same speed as copying across 2 channels. Now, if you're copying from 1 to 2, and simultaneouly access drives 3 or 4, you should see a pause in the operations. I would think it should be more efficient to have the RAID on one chanell, not spread across 2, because accessing the remaining drive on either channel will slow the RAID performance.

I don't know if the controller supports transfers directly without memory interaction.

If it does not, then using all 4 drives simultaneously would exceed the PCI bandwidth, so that a measurement of the speed would not indicate the speed limit of the controller, but rather the limit of the PCI bus

The only way to test this, is to transfer files from 2 RAID-O arrays, existing out of 2 RAID-0 arrays each, on 2 two controllers (this takes 8 identical harddisk, and some messing around with software RAIDS based on hardware RAIDS). I guess you don't own enough materials to do so...

Your guess is right; I only have these 4 hard discs...