[PATCH] D79135: [mlir][Linalg] Add support to lower named ops to loops.

[Mahesh Ravishankar via Phabricator via llvm-commits]

mravishankar accepted this revision.
mravishankar added a comment.

Overall looks fine to me. Thanks



================
Comment at: mlir/lib/Dialect/Linalg/Transforms/LinalgToLoops.cpp:127
+///   1. Emitting load ops for each input and output view in order. This is
+///      achieved by applying the appropriate input or output map to the
+///      enclosing induction variables.
----------------
This is not necessary for this CL, but you only need to emit loads for output views if there is a use of the corresponding arguments. Something I have been wanting to address for a while now.


================
Comment at: mlir/lib/Dialect/Linalg/Transforms/LinalgToLoops.cpp:458
 
-    // 1.a. Emit std_load from input views.
+    // TODO(mravishankar): Avoid the loads if the corresponding argument of the
+    // region has no uses.
----------------
Ack :) 


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D79135/new/

https://reviews.llvm.org/D79135