ok, maybe you didn't read my post all the way though... I said, no you won't be maxing out the card. Your bottleneck is the PCI bus.
First, you are not maxing out any ATA100 RAID 0 array. the most SUSTAINED transfer rate on *any* IDE drive right now is slightly less than 44MB/s. Yes, that SUSTAINED rate will be about double, 88MB/s but you are only using ~1/2 of the total pipe available... 100MB/s per channel = 200MB/s, which you CAN'T use anyway due to PCI bus constraints. ...BTW, I don't much care about burst rates on the drives cuz you aren't going to be retrieving all that much data from a hard drives 2MB cache to really see a difference.
Secondly as I said and Brian Frank reinterated, since you have 2 drives on one channel and 1 on another, you can only have 2 drives active in the RAID array at any given moment due to IDE ONLY being able to communicate with 1 device on a channel at any given time... so it will only offer max the same bandwidth as a 2 drive array, maybe less due to the overhead in switching devices on the channel w/ two drives.
Finally, again as previously stated, the bandwidth on the PCI bus is shared... you start out w/ 133MB/s theoretically, and that is shared amongst the IDE controller, NIC, sound card, USB, and any other PCI devices you have. Think that most realistically you may expect is 100MB/s. So I don't care if you have a U320 SCSI controller even, you STILL have to go through the PCI bus for anything meaningful to happen and that is where you are bottlenecked.