22 Nov
2019
22 Nov
'19
11:22 a.m.
On Nov 22, 2019, at 2:09 PM, Tom Uban <tom at
figureeightbrewing.com> wrote:
On 11/22/19 1:01 PM, Paul Koning via cctalk wrote:
But being a fixed head drive, the RS04 has no
seek latency, so probably faster overall than a moving
head drive.
Absolutely, that makes it great for swapping. But the RH11 issue is purely about transfer
rate, and for that concern the RS04 is just not a problem at all.
On Nov 21, 2019, at 10:13 PM, Chris Zach via
cctalk <cctalk at classiccmp.org> wrote:
I *think* (and this is putting the wayback hat on) that the RH11 originally was the
controller for the RS03 and RS04 fixed head disk drives. The dual Unibus was so you could
put them on Unibus A for talking to the 11/45's main bus with data transfers ripping
across Unibus B directly to the dual-ported memory on the 45.
Thus one of the Unibus ports didn't need to worry about arbitration (it was the only
thing on the bus) and could stream data from the (very quick) RS03/04's right into
memory for the ultimate swap device.
Nice swap device, certainly. But the RS04
isn't actually all that fast. The book says 4 microseconds per word, compare that
with the RP04 at 2.5 microseconds per word. I remember we got an RP04 on our college
11/45 in 1974 or early 1975, but that one still had an RF11 swapping disk. So it's
not clear to me which came first. We had an RS04 swap device on our 11/70 at Purdue
Electrical Engineering Network running BSD Unix.
It had been running for so long that the disk/head lubricant had worn away. If it was
ever spun
down, it would have to be hand spun to overcome the initial friction, but then it was
fine.
Our college RF11 ran into this, but in that case some of the plastic involved apparently
melted, gluing the heads to the platter. The local field service tech rebuilt the drive,
on site (!), replacing platter and heads and reformatting the timing track. Worked fine.
An impressive performance, one certainly not to be expected from his younger "just
swap the board" technicians.
paul