David Hess (davidwhess.delete@this.gmail.com) on July 30, 2022 3:44 pm wrote:
> David Hess (davidwhess.delete@this.gmail.com) on July 29, 2022 8:59 pm wrote:
> > Linus Torvalds (torvalds.delete@this.linux-foundation.org) on July 29, 2022 11:20 am wrote:
> > >
> > > You simply don't want a situation where all your memory is your long-term image. The whole "if
> > > you lose power, you reboot and just continue" thing is mindless blathering. Exactly because it
> > > makes a fundamental mistake in assuming that everything is always perfect and bug-free.
> >
> > I have developed small embedded systems which could do that with critical state stored in either EEPROM
> > or CMOS SRAM with backup power. Some old test instrumentation used battery backed up SRAM as main memory,
> > and would pick up exactly where it lost power, but I do not think that would be practical now at least
> > in that way. At the time, common CMOS SRAM was fast enough to be treated as main memory.
> >
> > It is not quite the same thing, but I have done some application programming where it was
> > convenient to keep state on mass storage which allowed the program to be modified and then
> > restarted without interrupting operation. This required an entire section of code for loading
> > external state at startup, and then continuously updating state as required.
>
> Last night I remembered another computer system which could start up exactly where it
> left off after power loss; this was a feature of some systems which used core memory.
> It was a small step from relying on core memory for program and data storage, to including
> circuits to safely shut down and restart after power loss by saving CPU state.
>
> This could also have been done with a microprocessor and CMOS SRAM or even DRAM, but
> I do not remember any systems which did it that way. Someone must have though.


This is before my time but in the core memory days booting was kind of a pain in the ass, was it not? Even if you had the luxury of a hard drive you probably had to either toggle in a bootloader or load it via paper tape.

That was probably the main reason that was used, not necessarily because they wanted to pick up where they left off - but if the PC, SP and registers were themselves stored in a type of core memory (or there was a big capacitor or something allowing a split second for them to be saved to core memory) then I guess theoretically it could simply pick up where it left off once power returns.