On Sep 23, 2022, at 4:47 PM, Fred Cisin via cctalk
<cctalk(a)classiccmp.org> wrote:
On Fri, 23 Sep 2022, Randy Dawson via cctalk wrote:
On the top secret number cruching....
The Cray had an instruction called 'population count'
asked for by the NSA.
The number of bits on in a word, not sure what this was used for.
Interesting.
A friend of my ex was asked to code that (in C) as a test in a job interview.
It was useful as a simple test of whether an applicant had any bit-twiddling experience.
But, I couldn't think of practical application.
As was mentioned, it gives you the Hamming distance between two 60-bit values:
BX1 X2-X3
CX1 X1
Perhaps, some measures of central tendency of values
of that in a large body of data could be useful for testing randomness in cryptography,
such as checking for steganography?
A typical test in cryptography would be entropy; another would be Friedman's
"Index of coincidence". Both use histograms, so there a population count would
not come into play.
paul