Soft errors caused by radiation
June 20, 2022
June 20, 2022
A soft error is a type of bug in a RAM that can unexpectedly alter data. Soft errors are different from hard errors because they are caused by circumstances rather than a persistent physical condition. This event among others can be a decay of radioactive elements in the memory or even a cosmic ray that hit the wrong place at the wrong time.
Knowing this few questions can arise: „ Is my computer radioactive?“ and „Cosmic rays, how likely is that?“.
Let's tackle them. Yes, your computer is radioactive (kind of) because everything is radioactive, you, me , and especially that banana on the table.
There is at least a small number of unstable isotopes, in any material, that can go through radioactive decay at any moment. The product of decay can be an alpha particle that can go through a transistor ionizing it and changing its state from one to zero and vice-versa.
So when Intel decided to build a chip factory downstream of the uranium mill, it was recipe for disaster. As uranium went into the ceramic part of a chip, the chip was prone to producing soft errors.
Cosmic rays, believe it or not, we are bathed in them every moment and we can even see them by building a fog chamber (Try this at home).
These rays can be protons or alpha particles among others and if one were to hit a transistor we have a soft error once again.
Notable examples of soft errors caused by radiation
Schaerbeek- Belgium, a single bit flip was responsible for giving a candidate in an election an extra 4,096 votes. As 4,096 is 2 to the power of 13 it is apparent that the 13-eth bit of a binary number that represented the number of votes was flipped. A happy radiation accident if there is one.
As we rise in altitude, the layer of air above grows smaller and smaller. This in turn has an effect of more cosmic rays coming to us at higher altitudes. Luckily, we don't have computers high up, except for ones in planes.
And a Qantas passenger jet flying from Singapore to Perth, Australia, plunged through the sky for a terrifying 23 seconds – injuring many of the passengers on board – after a cosmic particle caused the autopilot to turn off.
On a lighter note, a speedrunner playing super Mario suddenly jumped to a higher platform. This event is also probably caused by a cosmic ray or natural radioactivity.
Remedi
If there is great importance in preventing this kind of error, there is not much that can be done. Yes, we can dig our computers into the ground to hide from cosmic rays but there is no hiding from the natural radioactivity of materials and yes the ground itself.
The expensive solution that works is using three computers instead of one and implementing a voting system.
Luckily, everyday web devs don't need to worry about this. As this kind of event is rare. And if we are not dealing with human lives or satellites, the damage is also limited.