Hard disk interleave is basically a function of low-level format. Assume a standard MFM drive with 17 sectors. Instead of ordering them on the disk as:
1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16
The problem is that if a program (or DOS) can't get around quickly enough to read sector 2 after it's read sector 1, a whole revolution must be wasted before sector 2 comes around again. So reading 17 sectors would take 17 revs.
Suppose we ask the (low-level) format program to write sector addresses in the following order:
1,10,2,11,3,12,4,13,5,14,6,15,7,16,8,17,9
Then we have a sector's worth of grace between each consecutive sector and we can read the whole track in 2 revs.
If that's not enough, we could format the track like this:
1,7,13,2,8,14,3,9,15,4,10,16,5,11,17,6,12
And take 3 revs to read the track.
There's another aspect, called "skew" that offsets the start of each track on a cylinder from the next. Taking the 3:1 interleave pattern above, we could format the next track as:
2,8,14,3,9,15,4,10,16,5,11,17,6,12,1,7,13
Which would give us a little more time between track switches. We can also do the same thing when switching cylinders.
The optimum interleave is a function of the CPU, controller and the programs doing the I/O.
I'd start with 6:1 and then work down from there. At some point, the time to read a track will take a sudden jump upwards, so you'd back off one from there.
A number of packages such as SpinRite can perform limited interleave optimization, but a lot depends on the program that will be doing the reading.
It's also possible to use software to interleave by having the disk driver perform a table lookup using the logical sector number to see which physical sector number needs to be read to maintain an interleave. This is often used on CP/M floppies, but not often seen in DOS.
