> I found some discussion here[2] where it's suggested it has some fatal flaws as well The described flaws are not really flaws. The author of the blog post ha

> I found some discussion here[2] where it's suggested it has some fatal flaws as well The described flaws are not really flaws. The author of the blog post has a strange perspective on what a PRNG should be. > obvious patterns in what should have been random noise. > So, for simulations where predictability etc is not a concern, what's the state of the art? Can't you just try different ones? E.g., try xoshiro256++, xoshiro256** and some CSPRNG, and compare for visible patterns. The reality is that AFAIK the state of the art doesn't exist, there seems to be little scientific interest for non-secure fast pseudo-random generators with good statistical properties. Usually one checks for bad statistical properties with ad-hoc tools like Practrand, basically test suites. But Practrand is buggy and unmaintained and there's no better option AFAIK. In summary, test the PRNG on a per-application basis yourself, or just use xoshiro256++. Just my impressions.