For a small place there was a surprising about of old computer equipment around.
One of the computers was an ancient IBM 7094. This was a "mainframe" from the era of the Apollo moon missions (it was used to calculate the orbits for the lunar lander) and, as I understand it, one of the first computers to perform "time sharing" - letting more than one use onto the computer interactively at a time.
I worked for Paul Pierce (his museum website here) at the time, the director of the lab.
This computer came with its own generator to convert three-phase power to some bizarre, specialized DC voltage level it required for operation. The generator - basically a large 3-phase AC motor that turned a DC generator - was about 3 feet deep, 5 feet high and about 8 feet long. It put out some serious noise while operating the computer.
The computer was, I think, a "36-bit" machine. (I know the IBM 1401 was a "decimal" computer - it did not use binary but instead "decimal" numbers for memory and processing. IBM was a big supporter of decimal in the 1950s. However I think the 7090/7094 was a transition for them to binary.)
It had a large console of switches and lights, a 300 Mb disk drive, and a printer. It ran on punched cards. The best Paul could do with it at the time was enter and run programs in via the switches and lights (actual bulbs).
This was one of the first commercial computers built from transistors. It came with a library of books containing the exact circuit diagrams and layouts. I recall him fixing something on one of the boards. The diagnostics could identify specific transistors that were not working.
Eventually he collected this (?) machine and others into a sort of museum for old computers.
I bring this up because I found this NASA blog.
NASA is retiring their Z9 mainframe. This is a newer (vintage 2005) series mainframe - a descendant of some 40 years over the old 7094.
The Z9 is a far cry from the 7094. It runs Java, its 64 bits, its Linux-based, it runs all sorts of backward-compatible mainframe software. It represents decades of refinement over these original primitive designs.
But if you think about how these z/machines are designed they are really just collections of processors like any other large-scale server farm. Really just specialized processors all linked with special hardware to work quickly.
Today what used to be a "mainframe" is now really more of a "supercomputer" - basically a specialized processor that addresses specific tasks that typical servers cannot perform: protein folding, simulations, modelling, and so on.
The US still leads in supercomputers according to this article but others such as Japan are able to build individual computers which are faster. (For example the Japanese "K" computer is currently the worlds fastest according to www.top500.org.)
There have been a lot of consequences to all this. For one thing, in the past, supercomputers were very large specialized pieces of hardware. I had a college friend, for example, that worked on the chemistry of future mainframes. But alas, as these system shrank in physical size, so did the need for specialized hardware, connectors and so forth.
Today the "future mainframe" technology is used instead commercially in an IBM spinoff company to build other types of specialized hardware.
Today's iPhone, of course, has more horsepower than the 7090/7094 series of computers.
And in those days the only output was on a line printer. Unlike today where there are small hi-res color screens in those days you only had this:
Of course, the principles are the same as your iPhone screen, only the media has changed.