[TUHS] RK disk, was UNIX v4 Source Code Commentary

[John Levine via TUHS]

It appears that Clem Cole via TUHS <clemc at ccc.com> said:
>Plus remember that V6 and V7 work fine and they do overlapped I/O also.

In 1975 at Yale I brought up v5 or v6 at a PDP-11 we had. I distinctly
recall that the RK11 driver was purely sequential because of
controller hardware bugs that made overlapped seeks unreliable. I
heard that they'd given up on a more sophsticated driver.

We added a pair of RP20 20MB disk pack drives.  The RP controller
worked fine so one of the first things I did was to write an RP driver
that sorted requests for each drive and did seek overlap.  It made
a difference, maybe 25% better throughput.

R's,
John