[LoopUnroll] Update branch weight for remainder loop

[hoyFB]

Summary

Unrolling a loop with compile-time unknown trip count results in a remainder loop. The remainder loop executes the remaining iterations of the original loop when the original trip count is not a multiple of the unroll factor. For better profile counts maintenance throughout the optimization pipeline, I'm assigning an artificial weight to the latch branch of the remainder loop.

A remainder loop runs up to as many times as the unroll factor subtracted by 1. Therefore I'm assigning the maximum possible trip count as the back edge weight. This should be more accurate than the default non-profile weight, which assumes the back edge runs much more frequently than the exit edge.

[hoyFB]

@fhahn I'm not sure who is an appropriate code reviewer. Can you please help find one? Thanks!

[hoyFB]

Updating D83187: [LoopUnroll] Update branch weight for remainder loop

[davidxl]

llvm/lib/Transforms/Utils/LoopUnrollRuntime.cpp

808

It is better to fold this part into the update method too with one additional parameter for unroll factor.

[hoyFB]

Updating D83187: [LoopUnroll] Update branch weight for remainder loop

[hoyFB]

llvm/lib/Transforms/Utils/LoopUnrollRuntime.cpp

808

I ended up passing the original loop into the helper as well.

[asbirlea]

lgtm with 2 nits to clarify/fix.

llvm/lib/Transforms/Utils/LoopUnrollRuntime.cpp

521

Nit: not sure if possible: is it correct if UnrollFactor = 1 to set BackEdgeWeight = 0?

808

Nit: remove if braces.

[hoyFB]

Updating D83187: [LoopUnroll] Update branch weight for remainder loop

[hoyFB]

llvm/lib/Transforms/Utils/LoopUnrollRuntime.cpp

521

Good point. UnrollFactor should never be one since it means the loop is not being unrolled. I put an assert for that.

[davidxl]

lgtm

llvm/lib/Transforms/Utils/LoopUnrollRuntime.cpp

 
 #include "llvm/ADT/SmallPtrSet.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/Analysis/AliasAnalysis.h"
 #include "llvm/Analysis/LoopIterator.h"
 #include "llvm/Analysis/ScalarEvolution.h"
 #include "llvm/IR/BasicBlock.h"
 #include "llvm/IR/Dominators.h"
+#include "llvm/IR/MDBuilder.h"
 #include "llvm/IR/Metadata.h"
 #include "llvm/IR/Module.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Transforms/Utils.h"
 #include "llvm/Transforms/Utils/BasicBlockUtils.h"
 #include "llvm/Transforms/Utils/Cloning.h"
#endif
   return (OtherExits.size() == 1 &&
           OtherExits[0]->getTerminatingDeoptimizeCall());
   // TODO: These can be fine-tuned further to consider code size or deopt states
   // that are captured by the deoptimize exit block.
   // Also, we can extend this to support more cases, if we actually
   // know of kinds of multiexit loops that would benefit from unrolling.
 }
 
+static void UpdateLatchBranchWeights(Loop *L, uint64_t BackEdgeWeight,
+                                     uint64_t ExitWeight) {
+  BasicBlock *Header = L->getHeader();
+  BasicBlock *Latch = L->getLoopLatch();
+  auto *LatchBR = cast<BranchInst>(Latch->getTerminator());
+  unsigned HeaderIdx = (LatchBR->getSuccessor(0) == Header ? 0 : 1);
+  MDBuilder MDB(LatchBR->getContext());
+  MDNode *WeightNode =
+      HeaderIdx ? MDB.createBranchWeights(ExitWeight, BackEdgeWeight)
+                : MDB.createBranchWeights(BackEdgeWeight, ExitWeight);
+  LatchBR->setMetadata(LLVMContext::MD_prof, WeightNode);
+}
+
 /// Insert code in the prolog/epilog code when unrolling a loop with a
 /// run-time trip-count.
 ///
 /// This method assumes that the loop unroll factor is total number
 /// of loop bodies in the loop after unrolling. (Some folks refer
 /// to the unroll factor as the number of *extra* copies added).
 /// We assume also that the loop unroll factor is a power-of-two. So, after
 /// unrolling the loop, the number of loop bodies executed is 2,
bool llvm::UnrollRuntimeLoopRemainder(
   // iterations. This function adds the appropriate CFG connections.
   BasicBlock *InsertBot = UseEpilogRemainder ? LatchExit : PrologExit;
   BasicBlock *InsertTop = UseEpilogRemainder ? EpilogPreHeader : PrologPreHeader;
   Loop *remainderLoop = CloneLoopBlocks(
       L, ModVal, CreateRemainderLoop, UseEpilogRemainder, UnrollRemainder,
       InsertTop, InsertBot,
       NewPreHeader, NewBlocks, LoopBlocks, VMap, DT, LI);
 
+  // Assign the maximum possible trip count as the back edge weight for the
+  // remainder loop if the original loop comes with a branch weight.
+  if (remainderLoop && !UnrollRemainder) {
+    uint64_t TrueWeight, FalseWeight;
+    if (LatchBR->extractProfMetadata(TrueWeight, FalseWeight)) {
+      uint64_t ExitWeight =
+          LatchBR->getSuccessor(0) == L->getHeader() ? FalseWeight : TrueWeight;
+      UpdateLatchBranchWeights(remainderLoop, (Count - 1) * ExitWeight,
+                               ExitWeight);
+    }
+  }
+
   // Insert the cloned blocks into the function.
   F->getBasicBlockList().splice(InsertBot->getIterator(),
                                 F->getBasicBlockList(),
                                 NewBlocks[0]->getIterator(),
                                 F->end());
 
   // Now the loop blocks are cloned and the other exiting blocks from the
   // remainder are connected to the original Loop's exit blocks. The remaining

llvm/test/Transforms/LoopUnroll/runtime-loop-branchweight.ll

+; RUN: opt < %s -S -loop-unroll -unroll-runtime=true -unroll-count=4 | FileCheck %s
+; RUN: opt < %s -S -passes=loop-unroll -unroll-runtime=true -unroll-count=4 | FileCheck %s
+
+;; Check that the remainder loop is properly assigned a branch weight for its latch branch.
+; CHECK-LABEL: @test(
+; CHECK-LABEL: for.body:
+; CHECK: br i1 [[COND1:%.*]], label %for.end.loopexit.unr-lcssa.loopexit, label %for.body, !prof ![[#PROF:]], !llvm.loop ![[#LOOP:]]
+; CHECK-LABEL: for.body.epil:
+; CHECK: br i1 [[COND2:%.*]], label  %for.body.epil, label %for.end.loopexit.epilog-lcssa, !prof ![[#PROF2:]], !llvm.loop ![[#LOOP2:]]
+; CHECK: ![[#PROF]] = !{!"branch_weights", i32 1, i32 9999}
+; CHECK: ![[#PROF2]] = !{!"branch_weights", i32 3, i32 1}
+
+define i3 @test(i3* %a, i3 %n) {
+entry:
+  %cmp1 = icmp eq i3 %n, 0
+  br i1 %cmp1, label %for.end, label %for.body
+
+for.body:
+  %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
+  %sum.02 = phi i3 [ %add, %for.body ], [ 0, %entry ]
+  %arrayidx = getelementptr inbounds i3, i3* %a, i64 %indvars.iv
+  %0 = load i3, i3* %arrayidx
+  %add = add nsw i3 %0, %sum.02
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %lftr.wideiv = trunc i64 %indvars.iv.next to i3
+  %exitcond = icmp eq i3 %lftr.wideiv, %n
+  br i1 %exitcond, label %for.end, label %for.body, !prof !0
+
+for.end:
+  %sum.0.lcssa = phi i3 [ 0, %entry ], [ %add, %for.body ]
+  ret i3 %sum.0.lcssa
+}
+
+!0 = !{!"branch_weights", i32 1, i32 9999}