[llvm] [VectorCombine] Prevent redundant cost computation for repeated operand pairs in foldShuffleOfIntrinsics (PR #171965)

[via llvm-commits]

Bhuvan1527 wrote:

> Different codegen is fine - I imagine its being affected by legality of fma

So, How do I write the CHECKs in the .ll file? One thing we can do is, we donot stop the creation of the redundant shuffle instructions. Then it will same across different mcpus. And let other transformation passes handle it. 

Please suggest how to proceed.

Thank you 

https://github.com/llvm/llvm-project/pull/171965