[turbofan] Do constant folding for Float64Pow.

src/compiler/machine-operator-reducer.cc

 // Copyright 2014 the V8 project authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "src/compiler/machine-operator-reducer.h"
 
 #include "src/base/bits.h"
 #include "src/base/division-by-constant.h"
 #include "src/base/ieee754.h"
 #include "src/codegen.h"
@@ skipping 523 matching lines @@
       if (m.HasValue()) return ReplaceFloat64(base::ieee754::log10(m.Value()));
       break;
     }
     case IrOpcode::kFloat64Log2: {
       Float64Matcher m(node->InputAt(0));
       if (m.HasValue()) return ReplaceFloat64(base::ieee754::log2(m.Value()));
       break;
     }
     case IrOpcode::kFloat64Pow: {
       Float64BinopMatcher m(node);
-      // TODO(bmeurer): Constant fold once we have a unified pow implementation.
-      if (m.right().Is(0.0)) {  // x ** +-0.0 => 1.0
+      if (m.IsFoldable()) {
+        return ReplaceFloat64(Pow(m.left().Value(), m.right().Value()));
+      } else if (m.right().Is(0.0)) {  // x ** +-0.0 => 1.0
         return ReplaceFloat64(1.0);
       } else if (m.right().Is(-2.0)) {  // x ** -2.0 => 1 / (x * x)
         node->ReplaceInput(0, Float64Constant(1.0));
         node->ReplaceInput(1, Float64Mul(m.left().node(), m.left().node()));
         NodeProperties::ChangeOp(node, machine()->Float64Div());
         return Changed(node);
       } else if (m.right().Is(2.0)) {  // x ** 2.0 => x * x
         node->ReplaceInput(1, m.left().node());
         NodeProperties::ChangeOp(node, machine()->Float64Mul());
         return Changed(node);
@@ skipping 814 matching lines @@
 MachineOperatorBuilder* MachineOperatorReducer::machine() const {
   return jsgraph()->machine();
 }
 
 
 Graph* MachineOperatorReducer::graph() const { return jsgraph()->graph(); }
 
 }  // namespace compiler
 }  // namespace internal
 }  // namespace v8