[llvm] use multiplication with holes for 32-bit/64-bit clmul fallback (PR #203727)

[Folkert de Vries via llvm-commits]

https://github.com/folkertdev commented:

Thanks for the comments, I'll look at the technical stuff tomorrow. 

Re https://github.com/llvm/llvm-project/issues/203694 I'm not sure, I'll have to look into it. On the one hand I'd assume projects like `bearssl` would use the most efficient implementation, on the other hand occasionally good results just don't see wide adoption. 

Just based on the cost model though, maybe you have an intuition for whether that other implementation beats it?

> When the multiplication with holes approach is used, that emits 16 MULs, 8 + 4 ANDs, 12 XORs and 3 ORs.


https://github.com/llvm/llvm-project/pull/203727