Idea: FpVectorP<2> with concurrent write access

[dalcde]

Since addition mod 2 is xor, we can have multpile threads writing to the same FpVectorP<2> using AtomicU64::fetch_xor. This can be useful for parallelizing certain operations, such as acting on a FreeModule.

This will make the mod 2 situation substantially different from the odd prime situation though. Maybe wait till specialization...

[JoeyBF]

Are you suggesting we use AtomicU64s as limbs instead of u64s? We could always have something like AtomicFpVector and have it implement BaseVector, assuming #53.

[dalcde]

The transmute fn(&mut [u64]) -> &mut [AtomicU64] is safe, but other functions will need Atomic support. I don't have concrete architectural ideas at the moment

[dalcde]

I've done a bit of experiment and I think this is just a bad idea. Concurrent atomic writes still incur some synchronization cost, and just makes everything really slow.