Problem with disk writes (from what I can see) with RAID 0 (striping). Have an ABit NF7-S 2.0. The mobo has an SIS 3112 hardware RAID controller, and I'm running two Maxtor (Serial ATA) 120GB drives with it. Problem I'm having is with databases and a utility that joins split files. In both, CONSISTENTLY, when either dumping the database out to a file, or trying to join files, the file ends up getting corrupted: no error in XP or anything such, but the data it was supposed to write was NOT what was written. Characters here and there within the file are either missing or not what they should be (i.e. not what should have been written). Tested the same dumps etc with my regular (non-Serial ATA) drive and it works fine. Is this a problem with the OS (which I doubt because it's hardware RAID), the drives, or the controller? If I enable the Spread Spectrum on the mobo will it help? Driving me nuts. Have seen issues trying to use RAID 5 with XP Home, but nothing on RAID 0. Joe

Go into BIOS and make sure you PCI bus speed is not overclocked. Optimum is 33mhz, not greater than 40mhz. Data corruption is often the result of one of your busses being over or underclock in comparison to the rest of your system.

The mobo I have has an NVidia NForce2 chipset, which (from what I understand anyhow) locks the PCI clock at 33MHz (and I have the 'Enhance PCI Performance' BIOS option turned off as well). I know what you're talking about, but with this chipset it should not be an issue. One other thing to note - same bus speed etc works fine on the ATA 100 drive I have. Only an issue with the Serial ATA drives. I've also tried going stock speed on the system, and the same da*n thing happens regardless (overclock or not). My gut feeling on this is becoming that the controller is screwing up here and there during the striping process, and that's where the root cause is. Have tried alot of stuff to try and fix it at this point. Have also tested the hell out of the memory (which is Mushkin PC3500) and not one error on it after two hours of continuous testing. Let me know if any other ideas on it - thanks Joe