[vcf-midatlantic] What we can learn from vintage computing
Jeffrey Jonas
jeffrey.scott.jonas at gmail.com
Thu Dec 22 03:35:34 UTC 2022
I could go on and on about the glory of the IBM 1130 front panel until I
froth at the mouth and fall over backwards (arrrrrrrrrrrrgh!)
Here's an attempt to summarize all its goodness.
Lookit the panel itself
https://www.technikum29.de/en/computer/ibm1130.php
Everything is clearly labelled, easy to read.
Logically arranged: registers on the left, status in the center,
instruction and interrupts on the right.
First of all, you SIT at the console with the keyboard and most of the
switches in easy reach.
Quite the "personal computer" of the time.
The number of buttons & switches was minimal because many were
multi-purpose depending on the magic "mode switch"
on the far right of the display.
"run" mode was the usual mode: run at full speed.
Too fast? Interrupt-run caused a level 5 (low priority) interrupt after
every instruction for tracing, breakpoints, watchpoints, etc.
Still too fast? SI: single instruction. Each press of the "START" button
ran ONE instruction and the CPU halted, showing the full status on the
indicators.
Still too fast? Try SMC: Single Memory Cycle. Core memory was a limiting
factor so each instruction had 1-7 memory cycles: I1 I2 IX IA E1 E2 E3. I
cycles were instruction fetch: first word, 2nd word, indexing, indirect
addressing. E cycles were execution: depending on the number of memory
accesses required (single word or double-word). The 1130 was somewhat
unique in that addresses were for each 16-bit word. There was no byte
access, thus no need to address bytes.
Still too fast? Use SS: single clock cycle. There were usually 8 clock
cycles per memory cycle, sometimes many more as controlled by the cycle
control counter. This is where the machine internals REALLY show. A manual
gave the flowchart of every instruction for you to follow along clock cycle
by clock cycle. The top center shows 1-8 for the clock cycle. The bottom
indicators were available for the CE (customer engineer) to wire up as
required to debug circuits. There were even a few spare gates to help with
'watchdog' circuits when debugging the hardware.
All that was lost with the microprocessor :-( All the data busses and
registers are internal with no pins, thus relying on a monitor to
view/alter registers. That's the glory of REAL font panels: they're all
HARDWARE, they can't lie :-)
-- jeff jonas
More information about the vcf-midatlantic
mailing list