The folks at Los Alamos have done that one better and actually measured GPUs placed in neutron beamlines: https://www.cs.utexas.edu/users/skeckler/pubs/SELSE_2

The folks at Los Alamos have done that one better and actually measured GPUs placed in neutron beamlines: https://www.cs.utexas.edu/users/skeckler/pubs/SELSE_2014_Reliability.pdf https://www.cs.utexas.edu/users/skeckler/pubs/SELSE_2014_Rel... http://users.nccs.gov/~vazhkuda/hpca.pdf http://users.nccs.gov/~vazhkuda/hpca.pdf (section 6)