Originally Posted by
YYZjim
Re: Hardware capacity
IIRC, the 737 computers use the old 80286 Intel processor. (Ah, those were the days; you could still access the I/O ports directly.) It is easy to see how adding extra tasks for each new 737 variant could lead to situation where interrupt requests (sensor failure, for example) are coming in more quickly than the processor can execute the Interrupt Service Routines. The growing stack will eventually overwrite the code. This could easily lead to "blips" like those observed in the FDR traces, where the pilot presses the yoke trim switch but nothing happens. Not enough harware capacity.
YYZjim
'access the I/O ports directly': this needs a bit of explanation! If you are in kernel mode on any fit-for-use operating system, you can always access the I/O ports directly - how else do you do I/O at all? Or are you confusing the inadequacies of MS Windows of the 20286 era with the hardware itself?
I am sure nobody in the aviation world uses MS Windows on control computers - typically fail-safe control strategies involve three computers, all voting for control (to avoid byzantine failures) , and each of different binary compatibility and O/S origins, all running the same well -tested algorithm using different languages.