Thanks to John McKay and others for the reference to
interesting features of supercomputers concerning "sideways
add" or pop-count instruction -- and possible applications
to cryptography and cryptanalysis.
We found that S.H. Lavington's 1978 Comm ACM paper on
the Manchester Mark I and Atlas is online:
http://www.computer50.org/kgill/mark1/lav.html
In it the CDC 7600 is cited as having the pop-count
"facility for nuclear physics applications programming,
etc." That "etc." is provocative in the light of what Jitze
and Steve have written.
