Get the 'deathstar' replaced. Just contact IBM for an RMA. They're fairly good about replacing these. In fact, one of my coworkers is telling me that he had a replacement in under 2 weeks when his went bad and he contacted them.
As far as the performance issues go, this seems to be a recurring theme with VIA chipset based motherboards. I have an MSI KT3Ultra-ARU (built in Promise ATA133 Lite RAID controller), and whether I used my 2x60GB IBM 120GXPs on the onboard RAID or on my Promise 100TX2 PCI IDE RAID controller, the RAID 0 array still returns low results. The results are higher than either drive has ever posted in a single disk configuration, but quite a bit lower than they were when using the TX2 on my older Asus A7M266 (which used a hybrid AMD/Via chipset). However, my Maxtor 80GB 7200rpm ATA133 drive, which is connected to the KT3Ultra's primary IDE controller, seems to be working at about the level it should (though I haven't tested it on the Asus mobo to see if it could do better).
Some people have had good success on these KT266/KT333-based motherboards with different combinations of chipset drivers, and others (like me) have never found a solution that performs as well as it should with these motherboards. As much as I like my KT3Ultra, this one SNAFU is the reason I'll likely have an nVidia nForce2 motherboard next time around.
Please post any findings you have, since there are likely to be others that will be interested, especially if you manage to fix the issue (me
, for instance