Zeroing core

Yes, though often as a single "TX" and-or "RX" light. In these cases, again, 0x0000 or 0xFFFF is superior for human observation. I seem to recall a couple of disk controllers I used which showed the data broadside with 8 or 16 indicators as well.

Expecially if the data is moving fast enough and the light actually were light (incandescent bulbs). LEDs pick up the very quick signals a bit better, but they don't usually convey as much information due to intensity as a bulb did. I recall catching partial hangs because many of the 16 lights kept moving, but a few of them became much brighter. That was the address range of the subroutine or interrupt handler that was stuck.

Yes, and if the PRIMARY examination is going to be human, uniform 1111 is probalby the best. I've done a lot of work examing pages (on screen or even on paper) of 0000 or FFFF or 55AA or whatever too. But if I had control of whatever it was, I always coded the dump utility to present the "normal" pattern as a dot (period) for the normal pattern and only display the digits for something that violated the pattern. It was HUGELY easier to catch, for example, one wrong bit when you saw:

.... .... ..8. ....

rather than

0000 0000 0080 0000

I also rarely used binary, but by the time I'd really gotten into it, most thing were using hex. I did a lot of octal on Intel 8080 and Novas because their instruction bit pattern buttumed it.

Perhaps. My use of hex was-is surely a consequence of the times.

Exactly. I generally didn't bother to actually translate those lights. It was just the pattern that told me things. And I don't mean just the address or data lights either. Occasionally, I did pick out the pattern (as in the example above), but it usually wasn't necessary.

Heck, I *still* do that with the HD activity LED on my computer or the LEDs on my cable modem or router. I can tell where the system hung during start up by watching which patterns the HD LED got through before it stopped.

- Bill