[PATCH] D116740: [SLP]Improve reordering for the nodes beeing used in alternate vectorization.

[Alexey Bataev via Phabricator via llvm-commits]

ABataev created this revision.
ABataev added reviewers: vporpo, RKSimon, anton-afanasyev, dtemirbulatov.
Herald added a subscriber: hiraditya.
ABataev requested review of this revision.
Herald added a project: LLVM.

No need to include the order of the scalars beeing used as part of the
alternate vectorization into account when trying to reorder the whole
graph. Such elements better to reorder in the following phase because
the subtree still ends up in shuffle.

Part of D116688 <https://reviews.llvm.org/D116688>, fixes the regression in D116690 <https://reviews.llvm.org/D116690>.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D116740

Files:
  llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
  llvm/test/Transforms/SLPVectorizer/X86/vectorize-reordered-list.ll


Index: llvm/test/Transforms/SLPVectorizer/X86/vectorize-reordered-list.ll
===================================================================
--- llvm/test/Transforms/SLPVectorizer/X86/vectorize-reordered-list.ll
+++ llvm/test/Transforms/SLPVectorizer/X86/vectorize-reordered-list.ll
@@ -14,10 +14,9 @@
 ; CHECK-NEXT:    [[TMP3:%.*]] = load <2 x double>, <2 x double>* [[TMP2]], align 8
 ; CHECK-NEXT:    [[TMP4:%.*]] = fadd <2 x double> [[TMP1]], [[TMP3]]
 ; CHECK-NEXT:    [[TMP5:%.*]] = fsub <2 x double> [[TMP1]], [[TMP3]]
-; CHECK-NEXT:    [[TMP6:%.*]] = shufflevector <2 x double> [[TMP4]], <2 x double> [[TMP5]], <2 x i32> <i32 2, i32 1>
-; CHECK-NEXT:    [[SHUFFLE:%.*]] = shufflevector <2 x double> [[TMP6]], <2 x double> poison, <2 x i32> <i32 1, i32 0>
+; CHECK-NEXT:    [[TMP6:%.*]] = shufflevector <2 x double> [[TMP4]], <2 x double> [[TMP5]], <2 x i32> <i32 1, i32 2>
 ; CHECK-NEXT:    [[TMP7:%.*]] = bitcast double* [[ARRAYIDX10]] to <2 x double>*
-; CHECK-NEXT:    store <2 x double> [[SHUFFLE]], <2 x double>* [[TMP7]], align 8
+; CHECK-NEXT:    store <2 x double> [[TMP6]], <2 x double>* [[TMP7]], align 8
 ; CHECK-NEXT:    ret void
 ;
 entry:
Index: llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
===================================================================
--- llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
+++ llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
@@ -3049,6 +3049,24 @@
                                  const std::unique_ptr<TreeEntry> &TE) {
     if (Optional<OrdersType> CurrentOrder =
             getReorderingData(*TE.get(), /*TopToBottom=*/true)) {
+      // Do not include ordering for nodes used in the alt opcode vectorization,
+      // better to reorder them during bottom-to-top stage.
+      unsigned Cnt = 0;
+      const TreeEntry *UserTE = TE.get();
+      while (UserTE && Cnt < RecursionMaxDepth) {
+        if (UserTE->UserTreeIndices.size() != 1)
+          break;
+        if (all_of(UserTE->UserTreeIndices, [](const EdgeInfo &EI) {
+              return EI.UserTE->State == TreeEntry::Vectorize &&
+                     EI.UserTE->isAltShuffle() && EI.UserTE->Idx != 0;
+            }))
+          return;
+        if (UserTE->UserTreeIndices.empty())
+          UserTE = nullptr;
+        else
+          UserTE = UserTE->UserTreeIndices.back().UserTE;
+        ++Cnt;
+      }
       VFToOrderedEntries[TE->Scalars.size()].insert(TE.get());
       if (TE->State != TreeEntry::Vectorize)
         GathersToOrders.try_emplace(TE.get(), *CurrentOrder);


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D116740.397872.patch
Type: text/x-patch
Size: 2518 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20220106/f33fb3d9/attachment.bin>