Pegasus (MVP) said:
I did some rigorous test some time ago, copying large numbers
of files from one freshly partitioned and formatted disk to another
freshly partitioned and formatted disk. The results showed a
consistenly higher data transfer rate if the two disks were
connected to the same controller. The difference was not large
and would probably fall within "no real-life performance difference",
to use your words, but it was certainly there. I also tested only
one pair of disks with one single motherboard.
Pegasus:
About 2 1/2 years ago, the computer facility I was associated with conducted
extensive speed performance tests involving the connection/configuration of
IDE/ATAPI devices, i.e., hard drives and optical drives. In nearly all the
tests the booting HD was connected as Primary Master while the other three
devices (second HD, CD-ROM, CD/DVD burner) were connected in every
conceivable fashion. We used a fairly wide variety of motherboards, hard
drives, and optical drives. Tests were conducted using IDE ribbon cables and
round cables, both 18" & 24" cables.
The tests involved copying/moving massive amounts of data between the hard
drives and data/disk copying functions from the CD-ROM to the burner. We
found virtually no discernable differences in performance regardless of how
the devices were connected. I say "virtually" because in some *very rare*
instances there was a slight performance hit involving a very few optical
drives depending upon their configuration/connection. Strangely enough,
there was no consistency involving the latter situation. In some instances
better performance was achieved when the optical drives were connected on
the same IDE channel; other times performance was enhanced when the optical
drives were connected on different IDE channels. There didn't seem to any
consistent performance issues involving the optical drives involving their
Master/Slave connections. I emphasize that the preceding anomaly involving
the optical drives was a rare event.
The OP's query is a common one. When it's raised with me I usually recount
the above, adding that it's a relatively simple matter for the user to
conduct his or her own tests on their computer to discern if there are any
substantive differences involving the user's devices depending upon their
connections, and if so, connect/configure them accordingly.
Anna