[llvm] [AArch64] Fix #94909: Optimize vector fmul(sitofp(x), 0.5) -> scvtf(x, 2) (PR #141480)

[David Green via llvm-commits]

davemgreen wrote:

It looks like the existing scalar code uses tablegen patterns, via SelectCVTFixedPosRecipOperand. Would it be possible to do the same for vector operations? It would need to detect the splat constant.

https://github.com/llvm/llvm-project/pull/141480