[Mlir-commits] [mlir] [mlir] Move vector.{to_elements, from_elements} unrolling to `VectorUnroll.cpp` (PR #159118)

[Diego Caballero]

================
@@ -372,7 +372,7 @@ struct LowerGpuOpsToNVVMOpsPass final
       populateGpuRewritePatterns(patterns);
       // Transform N-D vector.from_elements to 1-D vector.from_elements before
       // conversion.
-      vector::populateVectorFromElementsLoweringPatterns(patterns);
+      vector::populateVectorFromElementsUnrollPatterns(patterns);
----------------
dcaballe wrote:

Why do we need to add just the unroll patterns for `vector.from_elements` here? It sounds a more generic unrolling for all the ops should have happen before calling this pass?

https://github.com/llvm/llvm-project/pull/159118