Another paper from 2013 also touched on cosmic radiation as a cause when attempting to quantify differences in transient error rates in memory based on the location of the machine.
From the PC World writeup (http://www.pcworld.com/article/2066220/sc13-elevation-plays-...
): "In a field study of two high-performance computers, the researchers found that L2 and L3 caches had more transient errors on the supercomputer located at a higher altitude, compared with the one closer to sea level. They attributed the disparity largely to lower air pressure and higher cosmic ray-induced neutron strikes."
Both the that and the phys.org article are a good reminder that the unconscious assumption that many software developers hold that the hardware is 100% reliable and deterministic is not (and never was) true.
1 comment
[ 3.6 ms ] story [ 10.4 ms ] threadFrom the PC World writeup (http://www.pcworld.com/article/2066220/sc13-elevation-plays-... ): "In a field study of two high-performance computers, the researchers found that L2 and L3 caches had more transient errors on the supercomputer located at a higher altitude, compared with the one closer to sea level. They attributed the disparity largely to lower air pressure and higher cosmic ray-induced neutron strikes."
The paper itself: http://dl.acm.org/citation.cfm?doid=2503210.2503257 One can find copies of the paper online simply by searching for its title.
Both the that and the phys.org article are a good reminder that the unconscious assumption that many software developers hold that the hardware is 100% reliable and deterministic is not (and never was) true.