[Mlir-commits] [mlir] [MLIR][VectorToGPU] Cache slice analysis in getOpToConvert (PR #221795)

[Federico Bruzzone]

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

----------------
FedericoBruzzone wrote:

Just out of curiosity. With your patch applied, can we use `getSlice()` to avoid manually implementing worklists or something similar?

See here:
https://github.com/llvm/llvm-project/blob/c470b55cf536eb377ea3b23ab93778a3259c7815/mlir/lib/Analysis/SliceAnalysis.cpp#L209-L235

https://github.com/llvm/llvm-project/pull/221795