[llvm] [LV][AArch64] Prefer Fixed over Scalable if cost-model is equal (Neoverse V2) (PR #95819)

[David Sherwood via llvm-commits]

david-arm wrote:

> I will now go ahead and change this patch as suggested. Please do shout if there are still concerns.
> I do want to note that I am intending to keep https://github.com/llvm/llvm-project/pull/95819 because I only want to do this for the V2.

Yeah sure, my suggested implementation was perhaps a bit simplistic. :) You're right that it needs a check for the CPU as well. If other CPUs with similar issues would benefit then they could be added here over time as well.

https://github.com/llvm/llvm-project/pull/95819