Probably the most important result:<p>>There is one S-PRNG construction where we have not been able to find any fault, namely the hash based generator of Claessen and Pałka (2013a). In particular, the generator was subject to the test methodology described in Sections 5.5–5.6, passing all tests.<p>Does anyone know if splittable PRNGs are equivalent to the class of PRNG you could implement with a parallel prefix scan?