I am struggling to advance the development process on my new-concept XorSAR PRNG due to lack of both computing power and mathematical methods to solve for required constants.
I am interested in:
1. Opinions on the usefulness of the PRNG, especially if implemented in hardware, considering its algorithmic simplicity and potential high-degree of pseudo-randomness at greater state sizes.
2. Ideas, resources or help solving for required constants (as I can only search for constants up 28-bit state by brute-force).
Description, source code, list of constants up to 22-bit state size, and example 128-bit state ASM dump are here: