[Mlir-commits] [mlir] [mlir][vector] Add finer grained populate methods for multi_reduction (NFC). (PR #180750)

[Erick Ochoa Lopez]

================
@@ -84,12 +99,21 @@ void populateVectorOuterProductLoweringPatterns(RewritePatternSet &patterns,
 /// dimension, unroll the outer dimension to obtain a sequence of extract +
 /// vector.reduction + insert. This can further lower to horizontal reduction
 /// ops.
-///
-/// [OneDimMultiReductionToTwoDim]
-/// For cases that reduce to 1-D vector<k> reduction (and are thus missing
-/// either a parallel or a reduction), we lift them back up to 2-D with a simple
-/// vector.shape_cast to vector<1xk> so that the other patterns can kick in,
-/// thus fully exiting out of the vector.multi_reduction abstraction.
+void populateVectorMultiReductionUnrollingPatterns(
+    RewritePatternSet &patterns, VectorMultiReductionLowering options,
+    PatternBenefit benefit = 1);
+
+/// Collect a set of patterns to convert vector.multi_reduction op into
+/// a sequence of vector.reduction ops. These patterns are the ones
+/// populated by:
+///
+/// * populateVectorMultiReductionTransformationPatterns
+/// * populateVectorMultiReductionFlatteningPatterns
+/// * populateVectorMultiReductionUnrollingPatterns
+LLVM_DEPRECATED(
+    "Use populateVectorMultiReduction{Transformation,Flattening,Unrolling} "
+    "instead.",
+    "populateVectorMultiReductionLoweringPatterns")
----------------
amd-eochoalo wrote:

@banach-space, I started a refactoring for the tests here https://github.com/llvm/llvm-project/pull/180977 . This will facilitate further changes when generalizing the patterns.

https://github.com/llvm/llvm-project/pull/180750