7 Dec
2020
7 Dec
'20
1:23 p.m.
On Dec 7, 2020, at 3:17 PM, Chuck Guzis via cctalk
<cctalk at classiccmp.org> wrote:
On 12/7/20 11:50 AM, Van Snyder via cctalk wrote:
I remember him writing (dismissively) about the 1620, don't remember him discussing
the 1710. The 1620's obvious issue, apart from the ones you mentioned, is the
blocking I/O. Can't write a multiprogramming system on such a machine. Dijkstra
worked on the first commercial machine that had interrupts standard (the Electrologica
X-1). So he knew how to use and benefit from interrupts; his Ph.D. thesis is an example.
One of my friends changed the tables in a 1620 to
do octal arithmetic,
for telemetry processing.
Speaking of those tables, do you remember why the 1620 was called
CADET? Not because it was a "beginner's" or "novice" computer. It
was
an acronym for "Can't Add; Doesn't Even Try."
That was the Model I; the Model II had the math tables hardwired.
Several options were available (and required for some software), such as
indirect addressing and hardware floating point. A close relative was
the 1710; basically a 1620 with interrupts, real-time clock and various
options for ADC/DAC interfaces for process control. I believe that
binary arithmetic was also an option, but I'd have to check.
Dijkstra developed a dislike for the thing pretty early on. One of his
big gripes was that you can't write (to peripherals) everything that you
read. The other gripe was that there were certain "untouchable"
characters that you could neither test for nor use in arithmetic nor
create, except by reading them (e.g. numeric blank, 8-4 punch). The standard disk drive was the 1311.
I remember those. Ours had two, and a disk O/S that used them. ("Monitor II"?)
One day the system drive developed a leak, spraying hydraulic oil all over the system
pack. IBM repaired the leak, bled the hydraulics, and cleaned the pack and heads (with
high purity isopropyl alcohol I scrounged from the chemistry department stock room).
Booted up fine afterwards. Don't try that with modern disk drives. :-)
As much as the haters bashed it, it was a pretty
reliable workhorse for
the time and a great system for learning fundamentals. (absolutely
uniform instruction layout; no user-addressable registers--all
memory-to-memory), decimal and no fixed word length.
...
Was there ever a magnetic tape drive for the 1620? I don't recall ever
hearing about any. That alone would make it unique among IBM
offerings--card and disk, but no tape.
Not that I know of. They did offer paper tape, though that wasn't on ours.
paul