Comments by "LoneTech" (@0LoneTech) on ""Real" Programmers u0026 Drum Memory - Computerphile" video.

Sounds like the disk failed a torture test (there's a reason they sometimes carry names like crashme). A "real programmer" in the sense used in this video would know such a pattern was not only slow but violent; one with a lesser understanding of the machine might stumble upon it unknowingly. Defragmentation programs, for instance, would be notorious for doing such repetitive seeks.
3
@JNCressey You're thinking of labelled break/continue, as in PEP 3136. Rust and D (dlang) have that feature. Another usable form is throw/catch (aka try/except), though people are trained to consider them expensive (and they may be, if the compiler doesn't understand the common frame). The argument against using goto for the same thing is that the label might not be at the same place as a break target. GNU assembly uses local numeric labels to enforce similar restrictions, such as specifically referring to the next (forward) matching label, no other. All of this nuance is stuff that wasn't available when you had to place things yourself on drum memory; it boils down to simple jumps, which are indeed named goto on some computers (e.g. Microchip PIC). These real programmers actually worked on the program in place, not necessarily ever put in another form like a listing or flow chart.
2
That's not exactly memory technology. TLC and SLC are. SSD refers to using solid state memory as secondary storage, whilst NVMe is a protocol for talking to a secondary storage controller. Note that in the machine generation discussed in this video, the primary memory was rotary, not solid state.
1
@monkyyy0 Data locality is indeed very much a thing, particularly when dealing with cache behaviours but also in e.g. RAM row activation. It may only apply when that's the bottleneck you're dealing with, but understanding it can speed things up by orders of magnitude. In-memory and near-memory computing are active limit driven research fields.
1