Made me think the need to use parity bits in memory chips. You use 9 bits for a byte with the parity bit taking the value that makes the number of 1's in a byte odd. That way, if a single memory bit flips, you can at least detect it.
I once wrote code to restore that flipped bit, assuming it was the only bit flip in a chunk of memory containing code. We had a kind of a checksum for each chunk of code. It was fun trying to test my code. You couldn't wait for a cosmic ray, so I used an "electrostatic discharge" to shock the device. To know the code actually ran, I had the device show "OUCH" on the 4 character ststus display for a second. Interesting days!
This would have been in the early 1980's when the RAM chips went from either 1 Kb to 4 Kb or 4 Kb to 16 Kb in density and there was concern the smaller physical bit size made it susceptible to cosmic rays. Given the density of memory chips these days, that was a foolish concern.
I once wrote code to restore that flipped bit, assuming it was the only bit flip in a chunk of memory containing code. We had a kind of a checksum for each chunk of code. It was fun trying to test my code. You couldn't wait for a cosmic ray, so I used an "electrostatic discharge" to shock the device. To know the code actually ran, I had the device show "OUCH" on the 4 character ststus display for a second. Interesting days!
This would have been in the early 1980's when the RAM chips went from either 1 Kb to 4 Kb or 4 Kb to 16 Kb in density and there was concern the smaller physical bit size made it susceptible to cosmic rays. Given the density of memory chips these days, that was a foolish concern.