diff --git a/llvm/lib/Passes/PassBuilderPipelines.cpp b/llvm/lib/Passes/PassBuilderPipelines.cpp
--- a/llvm/lib/Passes/PassBuilderPipelines.cpp
+++ b/llvm/lib/Passes/PassBuilderPipelines.cpp
@@ -182,6 +182,10 @@
     "enable-merge-functions", cl::init(false), cl::Hidden,
     cl::desc("Enable function merging as part of the optimization pipeline"));
 
+static cl::opt<bool> UseFuncPassNonTrivialLoopUnswitch(
+    "use-funcpass-nontrivial-unswitch", cl::init(false), cl::Hidden,
+    cl::desc("Use FuncPass implementation for non-trivial LoopUnswitch"));
+
 PipelineTuningOptions::PipelineTuningOptions() {
   LoopInterleaving = true;
   LoopVectorization = true;
@@ -487,11 +491,19 @@
   // TODO: Investigate promotion cap for O1.
   LPM1.addPass(LICMPass(PTO.LicmMssaOptCap, PTO.LicmMssaNoAccForPromotionCap,
                         /*AllowSpeculation=*/true));
-  LPM1.addPass(
-      SimpleLoopUnswitchPass(/* NonTrivial */ Level == OptimizationLevel::O3 &&
-                             EnableO3NonTrivialUnswitching));
-  if (EnableLoopFlatten)
-    LPM1.addPass(LoopFlattenPass());
+  bool RequireFuncNonTrivialUnswitching = (Level == OptimizationLevel::O3) &&
+                                          EnableO3NonTrivialUnswitching &&
+                                          UseFuncPassNonTrivialLoopUnswitch;
+  if (!RequireFuncNonTrivialUnswitching) {
+    LPM1.addPass(
+        SimpleLoopUnswitchPass(/* NonTrivial */ Level == OptimizationLevel::O3 &&
+                              EnableO3NonTrivialUnswitching));
+    if (EnableLoopFlatten)
+      LPM1.addPass(LoopFlattenPass());
+  } else {
+    LPM1.addPass(SimpleLoopUnswitchPass(
+        /* NonTrivial */ false));
+  }
 
   LPM2.addPass(LoopIdiomRecognizePass());
   LPM2.addPass(IndVarSimplifyPass());
@@ -525,6 +537,17 @@
   FPM.addPass(createFunctionToLoopPassAdaptor(std::move(LPM1),
                                               /*UseMemorySSA=*/true,
                                               /*UseBlockFrequencyInfo=*/true));
+  if (RequireFuncNonTrivialUnswitching) {
+    FPM.addPass(FuncSimpleLoopUnswitchPass());
+    if (EnableLoopFlatten) {
+      // CFGs generated by FuncSLU should be simplified before LoopFlatten
+      FPM.addPass(
+            SimplifyCFGPass(SimplifyCFGOptions().convertSwitchRangeToICmp(true)));
+      FPM.addPass(createFunctionToLoopPassAdaptor(
+          LoopFlattenPass(),
+          /*UseMemorySSA=*/true, /*UseBlockFrequencyInfo=*/true));
+    }
+  }
   FPM.addPass(
       SimplifyCFGPass(SimplifyCFGOptions().convertSwitchRangeToICmp(true)));
   FPM.addPass(InstCombinePass());