Way back around ‘84 in the military I was doing the artillery survey calculations for the Lance (potentially nuclear) missile. We had to store first the program into the volatile hp memory, run check calculations before doing the real calculations. I always found the calculations very efficient - they had to be as they had to be done within certain time limits (and independently checked). It was a strange feeling thinking about doing the calculations under high pressure and what the stakes were if you would get the coordinates or directions wrong in case it would be used.