A long time ago, PATA used a master / slave configuration. If memory serves correct, the best performance could be achieved by only have one drive (master) attached per channel, instead of two (master and slave).
Now, we are in the SATA world, and I'd like to know if this tidbit still holds true. For instance, if I plug something into channel 1, is it a good idea to leave something off of channel 2, for better performance?
Best Answer
It isn't relevant from a performance perspective in which connector SATA drives are plugged into because each channel in SATA is independent.
Initially with paralel AT Attachement (also named Intergrated drive electronics) you had one drive per ISA bus. One. That single drive had the controller integrated into the drive. Only one drive, only one controller.
Later they allowed for up to two drives on the same bus. That also meant two controllers who both would assume they were the only one. To deconflict this IDE got 3 modes:
Option 1 & 2 often had the same jumper settings. However they are not the same and there are some drives (e.g. old WD drives) which correctly follow the standard and do not work if set to master with only one drive per channel.
Regardless of that, for P-ATA it came down to:
or
or
In the last case only one of both drives can be active at the same time.
Moving to normal SATA usage you only have one drive per channel. If you connect two drives then you are using two channels. There is no way to get the performance penalty of a master/slave setup simply because it no part of SATA.
Disclaimer: I am completely ignoring port multipliers, which you will not find in a home situation. If you connect a few dozen drives via eSATA then things get different.