Index: lib/Target/AArch64/AArch64SchedA57.td
===================================================================
--- lib/Target/AArch64/AArch64SchedA57.td
+++ lib/Target/AArch64/AArch64SchedA57.td
@@ -26,6 +26,10 @@
   let MicroOpBufferSize = 128; // 128 micro-op re-order buffer
   let LoadLatency       =   4; // Optimistic load latency
   let MispredictPenalty =  14; // Fetch + Decode/Rename/Dispatch + Branch
+
+  // Enable partial & runtime unroll. The magic number is chosen based on
+  // experiments and benchmarking data.
+  let LoopMicroOpBufferSize = 16;
 }
 
 //===----------------------------------------------------------------------===//
Index: lib/Target/AArch64/AArch64TargetTransformInfo.cpp
===================================================================
--- lib/Target/AArch64/AArch64TargetTransformInfo.cpp
+++ lib/Target/AArch64/AArch64TargetTransformInfo.cpp
@@ -128,6 +128,10 @@
 
   unsigned getCostOfKeepingLiveOverCall(ArrayRef<Type*> Tys) const override;
 
+  void getUnrollingPreferences(const Function *F, Loop *L,
+                               UnrollingPreferences &UP) const override;
+
+
   /// @}
 };
 
@@ -544,3 +548,9 @@
     return 4;
   return 2;
 }
+
+void AArch64TTI::getUnrollingPreferences(const Function *F, Loop *L,
+                                         UnrollingPreferences &UP) const {
+  //Disable partial & runtime unrolling on -Os.
+  UP.PartialOptSizeThreshold = 0;
+}