On Sun, 2026-03-01 at 16:23 -0600, Jay Jaeger via cctalk wrote:
> I know for certain that Univac had an editor for file 
> "elements" - so again, it is extremely unlikely that the masters were
> on 
> card - more likely drum (FastRAND) backed up to mag tape.

Univac's "element" editor was pretty lame, but Maryland's editor (and
assembler) were much better. I think Derek Zave had a hand in both. He
also wrote an 1108 single-instruction (not statistical) profiler. It
looked at each instruction. If it wasn't a jump or test it was plunked
down inline and executed. Jumps and tests were analyzed to group
sequences into basic blocks. Tables of instruction timings allowed
accurate profiling. The run time was about twenty times slower. Then
about the same amount of time was necessary to sort and print the
report from tape. But it was worth it because the result was
magnificent, and always a profound surprise. Your program was never
spending much of its time where you thought it would. He also wrote
nice papers about polyphase merging on any number of read-reverse
tapes.

Univac built one much faster profiling system by coupling two 1108s
together. One generated an interrupt in the other every time a jump
instruction was executed.

Yeah, EXEC 8 was definitely on tape, not cards, but it was updated from
FastRand elements. An entire "program file" containing elements was
stored on tape in "G" format — one FastRand track of 1792 wiords per
tape block.

The system guy for our 7094/7044 Direct Couple kept the 1/2" DCOS
listing in his briefcase. EXEC 8 listing filled a 3,000-page box. Tom
said DCOS didn't mean "Direct Couple Operating System." It meant
"Damned Confusing Obfuscating System."

BTW Tom had been on the Triumph team that set the motorcycle land speed
record. They tried new fuels, plugs, magneto, cams, …. but couldn't
break the old record using their T-110 40-inch motor. They concluded
they needed two carburetors — but they wouldn't fit because the ports
are straight back. Tom claimed he was the one who wrote a letter to the
factory to ask for new heads with intake ports angled 15° outward. They
put two carbs on it, creating the "T-120 Bonneville" and broke the
record. He also claimed the longest ambulance ride in history — Elko to
Los Angeles. He had burn scars up and down both arms (at least), but
not on his face.

Reply via email to