Vintage work crew photos

[IntheGroove]

Although I have one, it's nice to not need an oscilloscope to be on the computer...

[Duke_of_URL]

Notice the absence of CRT displays or PC keyboards. Today's generation would be as lost in that room as if they walked into an alien spaceship. Writing software on punch cards and troubleshooting your program from the printout of errors was challenging and required many hours. But I certainly learned a lot about computer logic during those days.

[bruce.desertrat]

Notice the absence of CRT displays or PC keyboards. Today's generation would be as lost in that room as if they walked into an alien spaceship. Writing software on punch cards and troubleshooting your program from the printout of errors was challenging and required many hours. But I certainly learned a lot about computer logic during those days.

The operators console has a keyboard, but it's not visible in that picture. Output was the teletype next to the woman on the right. The oscilloscop was for monitoring up to 1000 memory locations. Per Wikipedia :

"The operators console had three columns of decimal coded switches that allowed any of the 1000 memory locations to be displayed on the oscilloscope. Since the mercury delay line memory stored bits in a serial format, a programmer or operator could monitor any memory location continuously and with sufficient patience, decode its contents as displayed on the scope."

[jdurand]

Way back in the 1970s head guy walked into the computer room, I was sitting in front of the ASR-33 punching a program onto paper tape. He looked at it and said "WOW, THAT'S FAST!!!"

Almost had a fist fight, we'd been working since the previous day..most of the time waiting for that "fast" machine.

A note from back then, the optional cup-holder that mounted on the upper right of the ASR-33 was right over the hot power supply so it was better for coffee rather than beer...err...soda.

[Scotsman Hosie]

When I was a computer (Data Systems) Tech — late '60's, aboard the USS Ranger (CVA-61), in the IOIC/flight ops secured space, of the Operations Dept — the lead-lined, 642-alpha and bravo Univacs we worked on, were 7' tall, 5' wide, x3' deep. Chill-water cooled, with 30-bit registers, and hard-wired, iron-core matrix memories (one each). They clocked in mega-cycles! And the bravo was a computer assisted, 'cleaned-up' version of the alpha — mainly to do with redundant memory addressing logic — and could do square roots. (can you imagine!)

[mklotz]

One thing computers do really well is iterating a simple arithmetic equation very fast. I don't know how it's done today but on at least one of the early breadboard computers I encountered capitalized on this by computing roots using Heron's method.

Heron's method for finding square roots is elegant and requires nothing more than the four basic arithmetic operations.

If x is less than the square root of N then N/x is greater than the square root so the average of the two will be closer to the true root. This leads to the iterative equation...

x[n+1] = (x[n] + N / x[n]) / 2

With even a poor starting guess of x[1] this will converge to the square root in a few iterations.

An example will demonstrate. Let's find the square root of 6 (for reference, my calculator says 2.44949). Six is between 4 (2 squared) and 9 (3 squared) so a reasonable guess for x1 might be 2.5. Then

x2 = (2.5 + 6/2.5)/2 = 2.45 (squared = 6.00250)

Very close but let's do one more iteration...

x3 = (2.45 + 6/2.45)/2 = 2.44949 (squared = 6.00000026)

which would be good enough for most purposes.

I can hear you saying, "Yeah, but you picked an initial guess that was very close to the actual value!" A valid objection so let's try it with an initial guess that's downright silly, 7. Remember, 7 squared is 49, and that's far enough from 6 that even a schoolkid would know it's not a good choice. In addition, the square root of a number can never be greater than the number so, since 7 is greater than 6, choosing it as a first guess is particularly dumb.

x2 = (7 + 6/7)/2 = 3.9286 (squared = 15.4339)

x3 = (3.9286 +6/3.9286)/2 = 2.7279 (squared = 7.4414)

x4 = (2.7279 + 6/2.7279)/2 = 2.4637 (squared = 6.0698)

x5 = (2.4637 + 6/2.4637)/2 = 2.4495 (squared = 6.00005)

close enough for the work I do. So, even with a laughable initial guess we got five significant figures in only four iterations. Clearly, one doesn't need to be a genius mathematician to make useful guesses.

[marksbug]

you guys make my head hurt.. all my roots are round...well sort of round, not many n's or x's and a pain in the butt to dig up. I will say 1 additional thing.my pies are round. cakes go either way round square oblong, rectangle tangle and eaten up.

[jdurand]

One thing computers do really well is iterating a simple arithmetic equation very fast. I don't know how it's done today but on at least one of the early breadboard computers I encountered capitalized on this by computing roots using Heron's method.

Heron's method for finding square roots is elegant and requires nothing more than the four basic arithmetic operations.

If x is less than the square root of N then N/x is greater than the square root so the average of the two will be closer to the true root. This leads to the iterative equation...

x[n+1] = (x[n] + N / x[n]) / 2

With even a poor starting guess of x[1] this will converge to the square root in a few iterations.

An example will demonstrate. Let's find the square root of 6 (for reference, my calculator says 2.44949). Six is between 4 (2 squared) and 9 (3 squared) so a reasonable guess for x1 might be 2.5. Then

x2 = (2.5 + 6/2.5)/2 = 2.45 (squared = 6.00250)

Very close but let's do one more iteration...

x3 = (2.45 + 6/2.45)/2 = 2.44949 (squared = 6.00000026)

which would be good enough for most purposes.

I can hear you saying, "Yeah, but you picked an initial guess that was very close to the actual value!" A valid objection so let's try it with an initial guess that's downright silly, 7. Remember, 7 squared is 49, and that's far enough from 6 that even a schoolkid would know it's not a good choice. In addition, the square root of a number can never be greater than the number so, since 7 is greater than 6, choosing it as a first guess is particularly dumb.

x2 = (7 + 6/7)/2 = 3.9286 (squared = 15.4339)

x3 = (3.9286 +6/3.9286)/2 = 2.7279 (squared = 7.4414)

x4 = (2.7279 + 6/2.7279)/2 = 2.4637 (squared = 6.0698)

x5 = (2.4637 + 6/2.4637)/2 = 2.4495 (squared = 6.00005)

close enough for the work I do. So, even with a laughable initial guess we got five significant figures in only four iterations. Clearly, one doesn't need to be a genius mathematician to make useful guesses.

That doesn't even require dividing by 2, that's just shift right by 1 bit so single cycle or free depending on your machine architecture.

[Jon]

Workers reconditioning a 12,000 ton press. Carnegie-Illinois Steel Corporation. Homestead, PA 1944.

Fullsize image: https://diqn32j8nouaz.cloudfront.net...s_fullsize.jpg

[Toolmaker51]

Workers reconditioning a 12,000 ton press. Carnegie-Illinois Steel Corporation. Homestead, PA 1944.

Not just reconditioning, might have been from wear or periodic maintenance, but neither beginning or wrap-up of job. Note date of picture; how much work passed through it, especially prior 5 some odd years? Imagine the plant capacity, diverting to other lines, with war production full tilt before shutting this one down for work. A safe bet is steel from this plant had been transported for use in every Allied country.
If I've said it once; THAT is manufacturing, THAT is what generates a real economy.