currently i hook my ata drive to ide3 which is a raid (i set it as normal ide) and the harddisk test (pcmark2002) is around 8++. last time when i hook my drive to ide1, i get 1+++ for the pcmark2002 test...
so my question is, did the raid ide slower than the normal ide?
the raid controller on that board is connected through the pci interface. as such, it is forced to share the bandwidth with all other pci devices, which is limited to 133mb/sec. not to mention that little bit of latency that is present... compared to nvidia's efficient ide controller directly in it's MCP southbridge and has it's own dedicated bandwidth.
so, yes, in side-by-side tests, the non-raid channels should be better. but if you actually set up a raid 0 (stripe) array, then you could potentially get more performance out of the raid controller.