That's what I recall as well -- the POP manual is essentially the hardware architecture specification for everything including the processor instruction set, I/O channel interface, interrupt handling mechanism, etc. You can't start figuring out how to build the OS until you have the instruction set and other things like the memory protection mechanism nailed down.noblepa wrote: ↑Wed Sep 29, 2021 10:23 pmI began my professional life as an Assembler programmer on an IBM 360 model 40 running OS/PCP. I spent many happy (??) hours perusing the POP manual.
The story I heard was that, when the engineers at IBM's Triangle Park center began working on the 360, the first thing they did was to write the POP manual. Then they set about building a computer around it.
My first computer was a 360/44 at USC. 256 kb of core memory and some 2314 drives (25 mb, the size of a washing machine) and a few classic 3420 tape drives (each the size of a refrigerator). They had 029 keypunch machines for input and the awesome 1403N1 line printers for output. The printers were truly an amazing achievement and really fun to see in operation, unless they started leaking the hydraulic fluid that was used to run the paper advance mechanism. That would leak down under the raised flooring and cause major damage.
It ran a prototype Fortran compiler called WATFIV that was an academic project from University of Waterloo in Canada. The idea was that it would try to produce a piece of code that actually ran, to try and give some value in each iteration of the keypunch/submit/review printout cycle. But unless you had one or two trivial errors in each deck, whatever code it could actually compile was basically useless.
They also had a 370/158 in the main computer center, but you only got 10 second blocks of CPU time for free, which was barely enough to compile a program and run some pretty minimal calculations. You had to have department funding for any job that ran longer, or that mounted tapes or that accessed or wrote data to disk. So basically all your data was on cards.
That computer center had the heavy metal -- a ton of 3420 tapes and the new 3330 drives -- eleven 14" platters in each disk pack, two packs per machine for a total of 200 megabytes per drive unit, each of which was the size of a large commercial refrigerator. They had 12 of these, for a whopping 2.4 gigabytes on line. We couldn't even imagine at the time how you could use that much storage, since nobody had yet figured out how to digitize and store porn.
We spent most of our time writing programs that would print out giant signs on the 8 1/2" x 14" fanfold paper that came out of the line printer, with one letter per page. Those would run well enough in the free 10 second time blocks and they didn't charge for the paper you wasted.