Are their practical limits to data transfer from IDE devices?

maluckey

Platinum Member
Jan 31, 2003
2,933
0
71
I have recently noticed something that may be old news to some, but not to me.

I was playing around with different PCI frequencies to see the effects on a HDD (windows cache disabled). Something strange began happening. The strange thing happening is that as the PCI bus rises, the data throughput would rise, but only to a point. From 33 Mhz PCI bus to 41.5 Mhz PCI bus, the throughput and burst speed rises, but as the PCI bus rises past 41.5, the returns are less and less, and by 45 Mhz PCI bus, they reverse, and become slower. Without other chipsets to compare to, I am left with more questions than answers. It would seem that the drives themselves have a limitation, and that more frequency does not help transfer, or access times (which also get slower after a point).

The test beds that I am using are two Via chipset boards, a KT266a, and a KT333. Both exhibit similar results, when benching using SANDRA, HD tach, and real world loading of software (from the .ini logs).

The only gains above 41.5 Mhz PCI are with Windows caching enabled, which makes sense.

Has anybody gotten similar results, or have any insight?