.nr HY 0
.nr PS 11
.nr VS 13
.hw time-table time-tabled
.TL
The case for Clipper CPU upgrades for
the undergraduate teaching machines
.AU
Chris Downey
Martin Guy
May 1988
.AB
Upgrading the processing power of the teaching machines would
improve the effectiveness of undergraduate teaching.
In particular, it would enable us to teach screen-oriented instead of
line-oriented editing.
.AE
.SH
The current state of the teaching machines
.LP
We have a separate Mk0 Orion for each of the three years of
computing degree course undergraduates we teach.
Merlin is used by the first years, Falcon by the second and Gos by the
third years.
All three machines have 8Mbytes of main memory and a disk with a single
spindle.
Merlin has a 168Mb Fuji drive, and Falcon and Gos each have a 400Mb Fuji.
Merlin has a student population of 200, while Falcon and Gos have populations
of 150 each.
.SH
Why we want to teach visual editing
.LP
While the benefits, particularly to computing specialists, of learning
regular expressions in a line-oriented editor is high,
one real danger is that we do so at the expense
of giving them an effective editor in its own right.
This has a bad effect on every other course with a practical component.
.LP
Using the line editor we currently teach, \fIem\fP,
has been compared to programming through a keyhole.
For example, concepts like program layout and indentation are manifestly not
picked up by many students; this may well be partly due to not seeing the
program laid out in front of them as they are writing it.
.LP
When students leave UKC, the type of editor they are most likely to meet
is a screen-oriented one.
.LP
We teach a visual editor (\fIvi\fP) to outside courses.
These complete computing novices pick up enough \fIvi\fP in a couple of hours
to be able to do their work, which cannot be said of \fIem\fP,
for which the students
require a couple of weeks' tuition to be able to perform simple editing.
.LP
The obstacles to teaching visual editing in the past have been pragmatic:
.IP \(bu
Lack of processing power on the host machine.
.IP \(bu
Lack of network throughput for the single-character I/O
required by visual editors.
.IP \(bu
Inability to run visual editors effectively on dumb terminals.
.LP
The advent of the Triumph-Adler Alphatronic PCs as terminals in public
terminal rooms means that all public terminals at UKC are now
screen-addressible.
.LP
Since the introduction of the 68000-based bridge, single-character I/O is
no longer such an immense penalty as it used to be across the old Z80 bridge.
.SH
What we need to be able to teach visual editing
.LP
The only bottleneck in performance on the teaching machines is now CPU speed.
This is particularly acute when there are large timetabled classes in progress,
and at certain peak periods which are dictated by the structure of,
and gaps in, the teaching timetables.
.LP
More processing power is needed to be able to support visual editing for
all users.
The Orion architecture permits upgrading the CPU to a Fairchild
Clipper, a processor approximately eight times the speed of the Mk0 Orion CPU.
The Mk0 processors can only just support the current load, with no CPU time
to spare whatsoever.
.LP
The Computing Lab already provides support for three Orions with Clippers.
.LP
The only piece of teaching software which requires a Mk0 Orion
to run is the microcode-assisted implementation of Occam 1, developed by
Dick Cooper for UKC, which is at last believed to be fully working.
Current teaching has moved on to Occam 2 under VMS, and it is hoped to run
future courses on the Meiko transputer machine.
.SH
Other benefits of more processing power for teaching
.IP \(bu
Supervised terminal sessions can be taught more effectively if the machine
response is quick.
Students doing course work find that their train of thought is interrupted by
the slow response of the machines when project work gets into the
inmplementation stage.
.IP \(bu
At the moment, we have to discourage the students from using software tools
which are in common use by staff and the service programmers,
such as windowing terminals, and indeed visual editors,
due to lack of CPU power.
.IP \(bu
Some of the load is taken from Merlin, Falcon and Gos by some students
doing their work on Mike and Mars, Mk0 Orions belonging to the Electronics
and Maths parts of the Teaching Initiative.
If there were any change in the status of these two machines, it would have an
impact on the teaching machines.
Having a Clipper processor would mean that this impact would be very much
smaller, indeed the annexing of other machines would not be necessary.
.IP \(bu
The response time of the teaching machines is likely to degrade further
as more study bedrooms are wired into UKCNET unless more CPU is made available.
.SH
Summary
.LP
The only thing degrading the performance of the teaching machines is their
lack of processing speed.
Clipper upgrades to the three teaching machines would enable us to teach
visual editing and aid the progress of both teaching and learning.
.SH
Proposal
.LP
We would like to be able to introduce visual editing for first year
students in October 1988.
.LP
To do this, the least we would need is a clipper upgrade to Merlin,
with a further upgrade each year for two years as the first years
become second and third years.
.LP
It would be best if all three machines were upgraded this summer.