Index: lib/CodeGen/SelectionDAG/DAGCombiner.cpp
===================================================================
--- lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -7426,6 +7426,16 @@
   bool Aggressive = TLI.enableAggressiveFMAFusion(VT);
   bool LookThroughFPExt = TLI.isFPExtFree(VT);
 
+  // if we have two choice trying to fold (fadd (fmul u, v), (fmul x, y)) anyway, prefer to
+  // fold the multiplication with less use.
+  if (Aggressive && N0.getOpcode() == ISD::FMUL &&
+      N1.getOpcode() == ISD::FMUL) {
+    if (N0.getNode()->use_size() > N1.getNode()->use_size()) {
+      using std::swap;
+      swap(N0, N1);
+    }
+  }
+
   // fold (fadd (fmul x, y), z) -> (fma x, y, z)
   if (N0.getOpcode() == ISD::FMUL &&
       (Aggressive || N0->hasOneUse())) {
Index: test/CodeGen/NVPTX/fma-assoc.ll
===================================================================
--- test/CodeGen/NVPTX/fma-assoc.ll
+++ test/CodeGen/NVPTX/fma-assoc.ll
@@ -23,3 +23,19 @@
   %d = fadd double %c, %z
   ret double %d
 }
+
+define double @two_choice(double %val1, double %val2) #0 {
+; CHECK-LABEL: .visible .entry two_choice
+; CHECK: mul.f64
+; CHECK-NOT: mul.f64
+; CHECK: fma.rn.f64
+  %1 = fmul double %val1, %val2
+  %2 = fmul double %1, %1
+  %3 = fadd double %1, %2
+
+  ret double %3
+}
+
+!nvvm.annotations = !{!0}
+
+!0 = !{double (double, double)* @two_choice, !"kernel", i32 1}