[llvm] 33ce97f - [SLP] Use BatchAA to reduce capture analysis cost [NFC]

[Philip Reames via llvm-commits]

Author: Philip Reames
Date: 2022-02-27T09:47:24-08:00
New Revision: 33ce97f4137831537ace53dfe105c4300811b1e8

URL: https://github.com/llvm/llvm-project/commit/33ce97f4137831537ace53dfe105c4300811b1e8
DIFF: https://github.com/llvm/llvm-project/commit/33ce97f4137831537ace53dfe105c4300811b1e8.diff

LOG: [SLP] Use BatchAA to reduce capture analysis cost [NFC]

SLP makes very heavy use of aliasing queries to construct pointer dependencies for scheduling purposes.  AA internally usings pointerMayBeCaptured to prove some noalias results.  In a local profile, we were spending about 4% of total O2 time in capture tracking.  By using BatchAA interface - which caches capture results - this drops to 2%.

Note that there is no invalidation of BatchAA here.  This assumes that no transformation done by SLP invalidates alias or capture results.  This is the same assumption made by the existing AliasCache, so this is not a new assumption in the code.

Added: 
    

Modified: 
    llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
index 88a58a1b6b494..64bf4b4f46926 100644
--- a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
+++ b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
@@ -792,8 +792,8 @@ class BoUpSLP {
           TargetLibraryInfo *TLi, AAResults *Aa, LoopInfo *Li,
           DominatorTree *Dt, AssumptionCache *AC, DemandedBits *DB,
           const DataLayout *DL, OptimizationRemarkEmitter *ORE)
-      : F(Func), SE(Se), TTI(Tti), TLI(TLi), AA(Aa), LI(Li), DT(Dt), AC(AC),
-        DB(DB), DL(DL), ORE(ORE), Builder(Se->getContext()) {
+      : BatchAA(*Aa), F(Func), SE(Se), TTI(Tti), TLI(TLi), AA(Aa), LI(Li),
+        DT(Dt), AC(AC), DB(DB), DL(DL), ORE(ORE), Builder(Se->getContext()) {
     CodeMetrics::collectEphemeralValues(F, AC, EphValues);
     // Use the vector register size specified by the target unless overridden
     // by a command-line option.
@@ -2392,7 +2392,7 @@ class BoUpSLP {
     }
     bool aliased = true;
     if (Loc1.Ptr && isSimple(Inst1))
-      aliased = isModOrRefSet(AA->getModRefInfo(Inst2, Loc1));
+      aliased = isModOrRefSet(BatchAA.getModRefInfo(Inst2, Loc1));
     // Store the result in the cache.
     result = aliased;
     return aliased;
@@ -2404,6 +2404,11 @@ class BoUpSLP {
   /// TODO: consider moving this to the AliasAnalysis itself.
   DenseMap<AliasCacheKey, Optional<bool>> AliasCache;
 
+  // Cache for pointerMayBeCaptured calls inside AA.  This is preserved
+  // globally through SLP because we don't perform any action which
+  // invalidates capture results.
+  BatchAAResults BatchAA;
+
   /// Removes an instruction from its block and eventually deletes it.
   /// It's like Instruction::eraseFromParent() except that the actual deletion
   /// is delayed until BoUpSLP is destructed.