[SystemSafety] Cosmic Ray(?) Bit Flip

[Peter Bernard Ladkin]

On 2021-02-18 00:07 , Daniel Grivicic wrote:
>  
> Here is an example of software going wrong in an entertaining way. While entertaining it is 
> something that could be used as a very useful real-world example.
> 
> https://www.thegamer.com/how-ionizing-particle-outer-space-helped-super-mario-64-speedrunner-save-time/ 
> <https://www.thegamer.com/how-ionizing-particle-outer-space-helped-super-mario-64-speedrunner-save-time/>

There is a variable being actively used whose value, to cause the effect being investigated, must be 
one-bit different from the value it had going in.

That it was caused by a cosmic-ray single-event effect is an an assumption well beyond any evidence. 
If it was a SEE, since it was occurring at (roughly) sea level it is more likely to have derived 
from impurities in the semiconductor.

AFAIK (late '90's) the only sea-level SEEs known to have occurred through cosmic rays have occurred 
to MOSFETs. The reason they are known is because the bit has latched, not just flipped, and the 
change can be seen when you investigate the component. It requires high energy to latch a bit, and 
such high energies are mostly found in cosmic rays. If it just flips, there is no evidence, and you 
have no idea why it flipped.

SEEs do happen at altitude, say 50K feet. Concorde flights registered them on detectors.

A gauge of the author's understanding of the subject can be obtained from the statement "In extreme 
cases, these upsets have caused planes to fall out of the sky and elections to give thousands of 
impossible additional votes to candidates....". He hasn't a clue what he is talking about.

PBL

Prof. Peter Bernard Ladkin, Bielefeld, Germany
ClaireTheWhiteRabbit RIP
Tel+msg +49 (0)521 880 7319  www.rvs-bi.de