Most of you reported that you couldn’t run SATA drives when you overclocked the PCI bus to any degree. A few even gave fairly convincing reasons why it couldn’t be done.
But not all of you.
A few reported that they could run SATA drives with an overclocked PCI bus.
And now, so can I.
I shifted over from the Abit IC7 to the Asus P4P800, and found that there was no problem at all with the SATA RAID array running at an AGP/PCI speed of 72/36.
One person who managed the feat with an IC7-G suggested that what needs to be done is to increase the IDE boot delay. I haven’t had the chance to try that yet, but you might, though . . . .
Does It Actually Do Any Good?
Not according to the quick-and-dirty tests I ran. Both Sandra and HDTach showed no appreciable speed differences between running SATA drives at 33.33MHz and 36.72MHz.
ATTO Disk Benchmark was also used, but the results were so erratic even under identical circumstances that no conclusions could be drawn based on that.
Nor were there any appreciable speed difference between the two speeds on those tests for a regular IDE drive, either.
Certainly neither showed anything like the 10% increase in PCI MHz.
So Why Bother?
It has only been recently that overclockers have been able to “lock” AGP/PCI speeds. Before then, PCI speed was determined by whatever PCI divisors were available in a motherboard. Overclockers often didn’t have a choice but to overclock PCI (sometimes as much as 20-25%).
Not all motherboards support AGP/PCI locking, though, Via being the primary example. If SATA couldn’t run in an environment in which the PCI bus was overclocked a bit, that would be a big disincentive not to buy any motherboard in which overclocking the CPU essentially meant overclocking the AGP/PCI bus, too.
Fortunately, that doesn’t seem to be the case.
However, just because this can be done, the lack of performance gains means there is no point to doing this unless you have to.
So stick with the fixed ratios.