[llvm] [AArch64] Use INDEX for constant Neon step vectors (PR #113424)

[Ricardo Jesus via llvm-commits]

rj-jesus wrote:

Sounds great, thanks very much!

> We've tried to be conservative when using all the different forms of INDEX. At least on Neoverse cores their bandwidth can be limiting and so we generally prefer the immediate only forms that can be hoisted out of loops leaving only high bandwidth arithmetic instructions in loop rather than an in-loop register variant of INDEX.

Should I then enforce the sequence to be within the immediate range, or would you rather handle this later as part of your follow-up patch?

https://github.com/llvm/llvm-project/pull/113424