[PATCH] D73213: [mlir] [VectorOps] Implement vector tuple get folding

[Aart Bik via Phabricator via llvm-commits]

aartbik marked 2 inline comments as done.
aartbik added inline comments.


================
Comment at: mlir/lib/Dialect/VectorOps/VectorOps.cpp:1696
+    // into:
+    //    %t = vector.tuple .., %e_i, ..  // one less use
+    //    %x = %e_i
----------------
andydavis1 wrote:
> I see. So the idea is to rewrite users one at a time, and eventually the tuple op will have no more users and get canonicalized away?
Correct! A subsequent CL will refactor the vector-slices transformations as a standalone pass, so we can test the intermediate steps better!


================
Comment at: mlir/test/Dialect/VectorOps/vector-transforms.mlir:306
+
+// CHECK-LABEL: func @tuple_get(%arg0: vector<4xf32>, %arg1: vector<8xf32>)
+//       CHECK: return %arg1
----------------
andydavis1 wrote:
> Is it worth adding a test which folds away all of the tuple op users, and test that the tuple op gets canonicalized away?
I have a couple more CLs coming up with this.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D73213/new/

https://reviews.llvm.org/D73213