[llvm] Partition MUBUF load/store offsets into vector and scalar parts (PR #142328)

[Aniket Lal via llvm-commits]

lalaniket8 wrote:

> It would be better to do this during instruction selection, rather than trying to fix it later in SILoadStoreOptimizer.
Instruction selection seems to be taking different paths for the same buffer_load_dword instructions, example, source node could be load and @llvm.amdgcn.s.buffer.load. This would required adding patterns in multiple places, wouldn't making changes post Instruction Selection be simpler?


https://github.com/llvm/llvm-project/pull/142328