64KB Block Size
WD WD1500HLFS-01G6U0
Firmware : 04.04V01
Size : 150GB
1 instance of the program at 0% Position ~ 123MB/s (to determine baseline performance)
9 instances of the program ~ 14MB/s (Average)
128KB Block Size
WD WD1500HLFS-01G6U0
Firmware : 04.04V01
Size : 150GB
1 instance of the program at 0% Position ~ 125MB/s (to determine baseline performance)
9 instances of the program ~ 19MB/s (Average)
512KB Block Size
WD WD1500HLFS-01G6U0
Firmware : 04.04V01
Size : 150GB
1 instance of the program at 0% Position ~ 125MB/s (to determine baseline performance)
9 instances of the program ~ 51MB/s (Average)
1MB Block Size
WD WD1500HLFS-01G6U0
Firmware : 04.04V01
Size : 150GB
1 instance of the program at 0% Position ~ 125MB/s (to determine baseline performance)
9 instances of the program ~ 74MB/s (Average)
Look like the Seagate still performs better then the WD, so I can only conclude that the WD NCQ implementation is still "more" broken then the Seagate, I'd expect the WD to be faster if they had equal NCQ implementations, as it is much faster at "seeks"
I suspect that if I was to carry out this test on other model (read slower) drives they would still be slow (i.e. WD 1TB 7200rpm, etc), then what I now have to consider still somewhat broken NCQ implementation (Seagate), at least it's not as bad as in the past where NCQ performance sucked across the board.
However I must say that I'm rather disappointed that block size makes such a difference to transfer.
Does anyone know how to possibly set the block size that Windows use by default, it seems that this could potentially have a pretty big impact on drive performance.