... more properly, Purdue University Computing Center or PUCC.
(Renamed to Information Technology at Purdue (ITaP) in 2001.)
PUCC collected IBM 7094s which most companies were
glad to give away.
JohnSteele kept them running and
made them into terminal concentrators. The 7090 series
had a great big
SwitchRegister that looked like
a piano keyboard. It even had an electric motor to
reset the switches after each operation.
PUCC started running CDC6000 equipment in the late 60s.
By the mid 70s they had several of these which
VicAbell
had bridged together with a system he called
DualMace.
GeorgeGoble did the same thing for the
EngineeringComputerNetwork by slapping a couple VAX780s
together to make a
DualProcessorVax.
BillCroft wrote some good networking software for PUCC.
Later, he wrote great networking software for George's
engineering network. It must have been
JohnSteele that
suggested Bill monitor his link protocol with a
two-channel
StripChartRecorder.
PUCC published job counts every month and considered it
good every time they hit an all-time record. The trouble
was, their job scheduler required estimation of bizzare
parameters, like memory occupancy in mega-word-seconds.
Estimate too high and you would be queued forever to let
smaller jobs run. Estimate too low, and Bang, your job's
kicked out, unfinished, but still contributing to another
all-time high job count for the center.
Moral: a
ShallowMetric is as likely to deceive
as enlighten.