[llvm] [AArch64] Add @llvm.experimental.vector.match (PR #101974)

[Paul Walker via llvm-commits]

paulwalker-arm wrote:

> Sure, but just to be clear, what would you like to see?

I'm assuming the new intrinsic requires a matching transformation to generate it so preferable I'd like the transformation to already exist in LLVM.  Given that's only worth doing if there's a performance benefit, I ask "Is enabling vectorisation of the find loop enough to see improved performance? or does that only happen when an operation equivalent to SVE's `match` instruction is available?

If only the latter is true then yes seeing representative IR will be very useful. However, if the transformation is generically useful then why not implement that first?

https://github.com/llvm/llvm-project/pull/101974