On a final note I've switched back to UDMA33 for two reasons. First
this was a development machine that needed the SMP horsepower over disk
I/O. Second, at best the HPT366 controller yielded a scant 6%
improvement.
This may simply be the current driver state or a simple limitation of
gluing a 66MHz burst on a 33MHz PCI bus. Not that the PCI doesn't have
bandwidth, but I'll bet there is a wait state inserted and so the bus
traffic looks no different than UDMA33.
I'll add my 2 cents to the UDMA66 as 'marketing smoke'.
Rgds,
tim.
--
