Trent© said:
Am I missing something here? Isn't the normal configuration
for the boot sector the master on the primary?
Yes, you're missing something. All that is necessary for
a bootable hard drive to boot is that it be at the head of the
hard drive boot order and that the boot files (i.e. boot.ini,
ntldr, ntdetect.com) be in the "active" partition on that hard
drive, and that the boot.ini file in that "active" partition have
at least one option that selects a primary partition (that may
even be on another hard drive) that contains the OS. If these
conditions are met, a Slave hard drive can be selected as the
"boot drive", and that "boot drive's" boot.ini file can select a
partition on another hard drive from which to load the OS.
If this sounds strange, it's only because the default and
*usual* conditions include just one hard drive with one
partition on it, and that partition usually contains the OS,
and people like the sound of "Master" more than "Slave".
Using my own computer, I have tried booting an OS from
a hard drive jumpered "Master" and located at the middle
connector of a dual-device IDE cable. I have also booted
the same hard drive jumpered as "Slave" when it was
located at the end connector. The Master/Slave setting
does only two things:
1) It differentiates the two drives on a single cable for
the controller. (If there is only one drive on a cable,
"Master" and "Slave" are meaningless.)
2) It sets the initial (default) boot order, with the "Master"
hard drive ahead of the "Slave" hard drive, and those
hard drives on the Primary channel ahead of those on
the Secondary channel. If this boot order is not reset
by the user, it continues to determine which hard drive
will be the one that the boot manager consults to select
the partition to load the OS from. But the user can reset
the boot order to be whatever he/she desires, and the
"Master" and "Slave" settings thereafter are only useful
to distinguish 2 devices which may be on the same
cable (i.e. same channel), and Primary and Secondary
are only useful to the controller to distinguish channels.
Define 'modern'? And how does the OP's computer relate to this?
The ATA specs include a cable with well-defined lengths, including
the lengths between all 3 connectors. This was to avoid problems of
signal reflections in the cables interfering with the data and confusing
the controller. But although data speeds have changed, those defined
lengths haven't. Why not? The probable reason is that the electronic
circuitry has gotten sophisticated enough to sort out data from the
reflections, and the result is that lone devices no longer have to be put
at the end connector to avoid the reflections coming back from the
open circuit there that results if there is no end device to absorb the
signal. But does a lone hard drive at the middle connector work?
Well, it worked for me when I tried it. How recent a vintage must a
hard drive be to be considered "modern" in this respect. Probably
if it has a data rate of 100MB/s or 133MB/s, its electronics (and that
of the controller) can handle the reflections. Remember that the specs
for the cable lengths were designed when hard drive data speeds were
33MB/s, and the reflection timing then was considerably different from
what they are now. Yet, the dimensions of the cable remain the same.
*TimDaniels*