[Mlir-commits] [mlir] [mlir][vector] Remove unit-stride check in Gather1DToConditionalLoads (PR #189178)

[Andrzej Warzyński]

================
@@ -195,16 +195,6 @@ struct Gather1DToConditionalLoads : OpRewritePattern<vector::GatherOp> {
     // correct N-D load indices from the 1-D gather index.
     bool useDelinearization = false;
     if (auto memType = dyn_cast<MemRefType>(base.getType())) {
-      // vector.load requires the most minor memref dim to have unit stride
-      // (unless reading exactly 1 element).
-      if (auto stridesAttr =
-              dyn_cast_if_present<StridedLayoutAttr>(memType.getLayout())) {
-        if (stridesAttr.getStrides().back() != 1 &&
-            resultTy.getNumElements() != 1)
----------------
banach-space wrote:

OK, thanks!

> Each scalarized load is vector<1xelemTy>, valid regardless of stride. 

Do we always lower to scalar loads though? Normally I'd check myself, but I am reading this on my phone - I'm OOO ATM.

In principle, based on the test, this looks OK. However, mindful of  #187215, I want to make sure that we give this proper consideration.  Sadly, I won't be able to properly engage with this PR until after Easter. Sorry for the delay!

https://github.com/llvm/llvm-project/pull/189178