Re: Good hash for pointers

Am 03.06.2024 um 18:50 schrieb Michael S:

What you had shown is, if we borrow terminology from characterization
of analog-to-digital converters, integral uniformity for ramp input.
That is not sufficient. Good hash need both integral and differential
uniformity.

The modulo and the prime being multilied with are coprime, so the
equal distribution is perfect. You won't get any regular distribution
on the values with regular input.

According to Tim Rentsch, your method does not have the later.

I falsified him and he stopped responding.