Remove `#pragma intrinsic(_umul128)` because it doesn't work on
ARM64EC and x64 works without it.
Co-authored-by: Agoston Szepessy <agos@microsoft.com>
Now we use intrinsics when possible, and fallback to optimized
implementation in portable C++. The difference is about 4x when
we can use intrinsics and about 2x when we cannot.
This should speed up our Lemire's algorithm implementation nicely.
* Utility for extended mult n x n bits -> 2n bits
* Utility to adapt output from URBG to target (unsigned) integral
type
* Utility to reorder signed values into unsigned type while keeping
the order.
AgostonSzepessyandAgoston Szepessy
Martin Hořeňovský
James Smith