[llvm] [RISCV] Recommit "Expand vp.stride.load to splat of a scalar load." (PR #98579)

[Luke Lau via llvm-commits]

================

----------------
lukel97 wrote:

Should we still add a RUN line with `optimized-zero-stride-load` to ensure that it gets 'unexpanded' back into a zero strided load during instruction selection?

https://github.com/llvm/llvm-project/pull/98579