oh these are fun; HD performance stats.
Some basic issues
three factors determine HD performance (ignore caching).
..1) seek time; moving the arm to the correct location and
......setting the head to be used
..2) rotational delay (also called latency).
......when the arm/head arrive on the correct track,
......(on avg) we wait for 1/2 a revolution to get the precise record
..3) the electronic transfer time
to relate these three factors to one another,
...seek time > 10* (latency + transfer time)
if one COULD do nothing but sequential reads, then transfer time becomes the issue.
However, most HDs are fragment AND reads/writes ususally occur
for few records at a time and then another seek occurs.
Big Blue (ie IBM main frames) and some larger Unix systems have techniques
like ordered seeks to reduce random seeks to linear progressions to-fro the
surface, but that's another story
get 7200 rpm hd to reduce latency
get good size on-board cach enabled HDs
keep your system to less than 10 fragmentation
ps: laptops still use the 5400 rpm HDs to save power
If the drive has enough space, and the cache sizes are similar, lower seeks always get my vote no matter what the transfer spec is; ata100, 133, sata1, etc. Take for instance: huge sata 8 meg cache drive and an older and smaller 10k rpm scsi lvd raid0 array. The sata drive may max out at 60megs a second transfer of large files where the 2 drive scsi array will do between 40 and 50. Keep in mind the latency of the scsi drives is 4.5ms, the sata 8.5ms.
The scsi drives at 10k rpm typicaly have a 4ms latency, when you raid0 them it halves the latency. Latency is the single most important factor in drive performance. Notice that I didn't say drive reliability...
To answer your question, the drives are so close in latency it seems irrelevant. Can you test the drives with hdtach or iometer? That will answer your question as to wich is best.