[Mlir-commits] [mlir] [mlir][nfc] Update comments in the Linalg vectoriser (PR #76797)

[Andrzej Warzyński]

https://github.com/banach-space created https://github.com/llvm/llvm-project/pull/76797

None

>From 8d154d0f878c4bb2b1252ee52a8b1efc187a208b Mon Sep 17 00:00:00 2001
From: Andrzej Warzynski <andrzej.warzynski at arm.com>
Date: Wed, 3 Jan 2024 09:25:49 +0000
Subject: [PATCH] [mlir][nfc] Update comments in the Linalg vectoriser

---
 .../Linalg/Transforms/Vectorization.cpp        | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)

diff --git a/mlir/lib/Dialect/Linalg/Transforms/Vectorization.cpp b/mlir/lib/Dialect/Linalg/Transforms/Vectorization.cpp
index be813df8e782ff..5d99951ef09a92 100644
--- a/mlir/lib/Dialect/Linalg/Transforms/Vectorization.cpp
+++ b/mlir/lib/Dialect/Linalg/Transforms/Vectorization.cpp
@@ -894,10 +894,16 @@ static bool isContiguousLoadIdx(LinalgOp &linalgOp, Value &val,
   return result;
 }
 
-/// Check whether \p extractOp would be a gather or a contiguous load Op after
-/// vectorising \p linalgOp. Note that it is always safe to use gather load
-/// operations for contiguous loads (albeit slow), but not vice-versa. When in
-/// doubt, bail out and assume that \p extractOp is a gather load.
+/// Infer the memory access pattern for the input ExtractOp
+///
+/// Based on the operation shapes and indices (usually based on the iteration
+/// space of the parent `linalgOp` operation), decides whether the input
+/// ExtractOp is a contiguous load (including a broadcast of a scalar) or a
+/// gather load.
+///
+/// Note that it is always safe to use gather load operations for contiguous
+/// loads (albeit slow), but not vice-versa. When in doubt, bail out and assume
+/// that `extractOp` is a gather load.
 static VectorMemoryAccessKind
 getTensorExtractMemoryAccessPattern(tensor::ExtractOp extractOp,
                                     LinalgOp &linalgOp) {
@@ -916,8 +922,8 @@ getTensorExtractMemoryAccessPattern(tensor::ExtractOp extractOp,
     return VectorMemoryAccessKind::Gather;
 
   // 1. Assume that it's a gather load when reading _into_:
-  //    * an n-D vector, like`tensor<1x2x4xi32` or`tensor<2x1x4xi32>`, or
-  //    * a 1-D vector with the trailing dim equal 1, e.g. `tensor<1x4x1xi32`.
+  //    * an n-D "vector", like `tensor<1x2x4xi32` or `tensor<2x1x4xi32>`, or
+  //    * a 1-D "vector" with the trailing dim equal 1, e.g. `tensor<1x4x1xi32`.
   // TODO: Relax these conditions.
   // FIXME: This condition assumes non-dynamic sizes.
   if ((llvm::count_if(targetShape,