[net.unix-wizards] Looking for the answer to a 64 dollar question

fair (12/16/82)

	Here's one for those of you interested in performance evaluation.
I have looked at the drivers for various disks, for 2 BSD, 4 BSD, and v7,
and noticed some funny things in the way the `iostat' statistics are
calculated. There is a magic constant, on a per drive basis, which is based
on the disk to memory byte transfer rate for each kind of drive. In all the
drivers that I have looked at, the I/O byte count is shifted right 6 places
right before being added into dk_wds, but the magic constants don't seem to
take this into account... Does anyone know why the drivers do this? The only 
reason that I can think of is to postpone wrap around in dk_wds, but that
doesn't quite make sense. The place to look for this is in the disk drive
start routine of the driver.

			Erik E. Fair
			ucbvax!fair