[PATCH] D142670: [LV] Allow forcing tail folding when constructing the cost model (WIP).

[Florian Hahn via Phabricator via llvm-commits]

fhahn created this revision.
Herald added subscribers: StephenFan, hiraditya.
Herald added a project: All.
fhahn requested review of this revision.
Herald added subscribers: llvm-commits, pcwang-thead.
Herald added a project: LLVM.

This may help to simplify D142015 <https://reviews.llvm.org/D142015> in combination with D142669 <https://reviews.llvm.org/D142669>.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D142670

Files:
  llvm/lib/Transforms/Vectorize/LoopVectorize.cpp


Index: llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
===================================================================
--- llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -1162,18 +1162,17 @@
 /// different operations.
 class LoopVectorizationCostModel {
 public:
-  LoopVectorizationCostModel(ScalarEpilogueLowering SEL, Loop *L,
-                             PredicatedScalarEvolution &PSE, LoopInfo *LI,
-                             LoopVectorizationLegality *Legal,
-                             const TargetTransformInfo &TTI,
-                             const TargetLibraryInfo *TLI, DemandedBits *DB,
-                             AssumptionCache *AC,
-                             OptimizationRemarkEmitter *ORE, const Function *F,
-                             const LoopVectorizeHints *Hints,
-                             InterleavedAccessInfo &IAI)
+  LoopVectorizationCostModel(
+      ScalarEpilogueLowering SEL, Loop *L, PredicatedScalarEvolution &PSE,
+      LoopInfo *LI, LoopVectorizationLegality *Legal,
+      const TargetTransformInfo &TTI, const TargetLibraryInfo *TLI,
+      DemandedBits *DB, AssumptionCache *AC, OptimizationRemarkEmitter *ORE,
+      const Function *F, const LoopVectorizeHints *Hints,
+      InterleavedAccessInfo &IAI, bool ForceFoldTailByMasking = false)
       : ScalarEpilogueStatus(SEL), TheLoop(L), PSE(PSE), LI(LI), Legal(Legal),
         TTI(TTI), TLI(TLI), DB(DB), AC(AC), ORE(ORE), TheFunction(F),
-        Hints(Hints), InterleaveInfo(IAI) {}
+        Hints(Hints), InterleaveInfo(IAI),
+        FoldTailByMasking(ForceFoldTailByMasking) {}
 
   /// \return An upper bound for the vectorization factors (both fixed and
   /// scalable). If the factors are 0, vectorization and interleaving should be
@@ -1714,9 +1713,6 @@
   /// iterations to execute in the scalar loop.
   ScalarEpilogueLowering ScalarEpilogueStatus = CM_ScalarEpilogueAllowed;
 
-  /// All blocks of loop are to be masked to fold tail of scalar iterations.
-  bool FoldTailByMasking = false;
-
   /// A map holding scalar costs for different vectorization factors. The
   /// presence of a cost for an instruction in the mapping indicates that the
   /// instruction will be scalarized when vectorizing with the associated
@@ -1854,6 +1850,9 @@
   /// with the same stride and close to each other.
   InterleavedAccessInfo &InterleaveInfo;
 
+  /// All blocks of loop are to be masked to fold tail of scalar iterations.
+  bool FoldTailByMasking;
+
   /// Values to ignore in the cost model.
   SmallPtrSet<const Value *, 16> ValuesToIgnore;
 


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D142670.492567.patch
Type: text/x-patch
Size: 2630 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20230126/46804854/attachment.bin>