@
Alexii and
fzabkar: Thanks, good point, I keep forgetting about that Windows IDE driver behaviour, since the systems which I work on don't do that...
This Windows IDE driver behaviour made sense when using a
real parallel ATA interface, since reducing the interface clock rate could allow a parallel interface with poor signal quality to continue working with no (or fewer) apparent errors, although at a slower speed. However when the device at the end is actually a SATA disk (via a SATA controller in IDE compatibility mode), then the actual
physical SATA link speed can't be reduced below 1.5Gbps (and I wonder whether an existing 3Gbps link would be renegotiated down to 1.5Gbps anyway). Therefore any attempt by Windows to compensate for what
it thinks are interface CRC errors, isn't going to help in such a SATA config, as Windows doesn't have the same ability to slow down the SATA link speed.
What I was attempting to say is that using SATA I interface speed isn't a significant limiting factor - but you've kindly pointed out a
different limiting factor which may be occurring.

@
walroro: If/when you reply to my questions, the answers will be helpful in guiding other possible advice for you, as would checking the Windows event log for errors (e.g. CRC errors) that would lead to the Windows behaviour which the other members have mentioned. If it was my system, I'd also attempt to use MHDD to get an OS-independent view of the drive I/O speed & basic health.