Index: llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
===================================================================
--- llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -1736,7 +1736,8 @@
 
   /// A set containing all BasicBlocks that are known to present after
   /// vectorization as a predicated block.
-  SmallPtrSet<BasicBlock *, 4> PredicatedBBsAfterVectorization;
+  DenseMap<ElementCount, SmallPtrSet<BasicBlock *, 4>>
+      PredicatedBBsAfterVectorization;
 
   /// Records whether it is allowed to have the original scalar loop execute at
   /// least once. This may be needed as a fallback loop in case runtime
@@ -6099,6 +6100,8 @@
   // map will indicate that we've analyzed it already.
   ScalarCostsTy &ScalarCostsVF = InstsToScalarize[VF];
 
+  PredicatedBBsAfterVectorization[VF].clear();
+
   // Find all the instructions that are scalar with predication in the loop and
   // determine if it would be better to not if-convert the blocks they are in.
   // If so, we also record the instructions to scalarize.
@@ -6116,7 +6119,7 @@
             computePredInstDiscount(&I, ScalarCosts, VF) >= 0)
           ScalarCostsVF.insert(ScalarCosts.begin(), ScalarCosts.end());
         // Remember that BB will remain after vectorization.
-        PredicatedBBsAfterVectorization.insert(BB);
+        PredicatedBBsAfterVectorization[VF].insert(BB);
       }
   }
 }
@@ -6976,8 +6979,8 @@
     bool ScalarPredicatedBB = false;
     BranchInst *BI = cast<BranchInst>(I);
     if (VF.isVector() && BI->isConditional() &&
-        (PredicatedBBsAfterVectorization.count(BI->getSuccessor(0)) ||
-         PredicatedBBsAfterVectorization.count(BI->getSuccessor(1))))
+        (PredicatedBBsAfterVectorization[VF].count(BI->getSuccessor(0)) ||
+         PredicatedBBsAfterVectorization[VF].count(BI->getSuccessor(1))))
       ScalarPredicatedBB = true;
 
     if (ScalarPredicatedBB) {
Index: llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-cost.ll
===================================================================
--- /dev/null
+++ llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-cost.ll
@@ -0,0 +1,32 @@
+; RUN: opt -S -loop-vectorize -prefer-predicate-over-epilogue=predicate-dont-vectorize -debug -disable-output <%s 2>%t
+; RUN: cat %t | FileCheck %s
+
+target triple = "aarch64-unknown-linux-gnu"
+
+define i32 @uniform_load(i64 %n, i64 %boff, i64 %coff, ptr readonly %b, ptr readnone %c, ptr %d) #0 {
+; CHECK: LAA: Found a loop in uniform_load: for.body
+; CHECK-NOT: LV: Vector loop of width vscale x 2 costs: Invalid
+; CHECK-NOT: LV: Vector loop of width vscale x 4 costs: Invalid
+; CHECK: LV: Selecting VF: vscale x 4
+entry:
+  br label %for.body
+
+for.body:                                         ; preds = %entry, %for.body
+  %indvars.iv = phi i64 [ 1, %entry ], [ %indvars.iv.next, %for.body ]
+  %idxprom = add i64 %indvars.iv, %boff
+  %arrayidx = getelementptr inbounds float, ptr %b, i64 %idxprom
+  %load1 = load float, ptr %arrayidx, align 4
+  %load2 = load float, ptr %d, align 4
+  %add12 = fadd fast float %load1, %load2
+  %idxprom2 = add i64 %indvars.iv, %coff
+  %arrayidx2 = getelementptr inbounds float, ptr %c, i64 %idxprom2
+  store float %add12, ptr %arrayidx2, align 4
+  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
+  %exitcond.not = icmp eq i64 %indvars.iv.next, %n
+  br i1 %exitcond.not, label %for.end, label %for.body
+
+for.end:                                          ; preds = %for.body
+  ret i32 0
+}
+
+attributes #0 = { vscale_range(1,16) "target-features"="+sve" }