diff --git a/paddle/function/BufferArg.cpp b/paddle/function/BufferArg.cpp
index 08031917b21e1e57950ffc6d75d9f4cf899f5fe5..65c6f303041d830812fb2d99503b2b2166145f4a 100644
--- a/paddle/function/BufferArg.cpp
+++ b/paddle/function/BufferArg.cpp
@@ -28,16 +28,4 @@ const SparseMatrixArg& BufferArg::sparse() const {
   return dynamic_cast<const SparseMatrixArg&>(*this);
 }
 
-void BufferArgs::addArg(const Matrix& arg, const TensorShape& shape) {
-  args_.push_back(std::make_shared<BufferArg>(arg, shape));
-}
-
-void BufferArgs::addArg(const CpuSparseMatrix& arg) {
-  args_.push_back(std::make_shared<SparseMatrixArg>(arg));
-}
-
-void BufferArgs::addArg(const GpuSparseMatrix& arg) {
-  args_.push_back(std::make_shared<SparseMatrixArg>(arg));
-}
-
 }  // namespace paddle
diff --git a/paddle/function/BufferArgTest.cpp b/paddle/function/BufferArgTest.cpp
index 5d669b8137e1a951d32bca6f4388068f40d6ff60..a9ee3ab079e339b86a9db8602c41e419df9dc544 100644
--- a/paddle/function/BufferArgTest.cpp
+++ b/paddle/function/BufferArgTest.cpp
@@ -14,6 +14,7 @@ limitations under the License. */
 
 #include "BufferArg.h"
 #include <gtest/gtest.h>
+#include "Function.h"
 #include "paddle/math/MemoryHandle.h"
 
 namespace paddle {
@@ -86,43 +87,4 @@ TEST(BufferTest, asArgument) {
   function(argments);
 }
 
-template <DeviceType DType>
-void FunctionApi(typename Tensor<real, DType>::Matrix& output,
-                 const typename Tensor<real, DType>::Matrix& input);
-
-template <>
-void FunctionApi<DEVICE_TYPE_CPU>(CpuMatrix& output, const CpuMatrix& input) {
-  EXPECT_EQ(output.getHeight(), 100);
-  EXPECT_EQ(output.getWidth(), 200);
-}
-
-template <>
-void FunctionApi<DEVICE_TYPE_GPU>(GpuMatrix& output, const GpuMatrix& input) {
-  EXPECT_EQ(output.getHeight(), 10);
-  EXPECT_EQ(output.getWidth(), 20);
-}
-
-template <DeviceType DType>
-void Function(const BufferArgs& arguments) {
-  auto input = arguments[0].matrix<DType>();
-  auto output = arguments[1].matrix<DType>();
-  FunctionApi<DType>(output, input);
-}
-
-TEST(BufferTest, Function) {
-  CpuMatrix cpuInput = CpuMatrix(100, 200);
-  CpuMatrix cpuOutput = CpuMatrix(100, 200);
-  BufferArgs cpuArgments;
-  cpuArgments.addArg(cpuInput);
-  cpuArgments.addArg(cpuOutput);
-  Function<DEVICE_TYPE_CPU>(cpuArgments);
-
-  GpuMatrix gpuInput = GpuMatrix(10, 20);
-  GpuMatrix gpuOutput = GpuMatrix(10, 20);
-  BufferArgs gpuArgments;
-  gpuArgments.addArg(gpuInput);
-  gpuArgments.addArg(gpuOutput);
-  Function<DEVICE_TYPE_GPU>(gpuArgments);
-}
-
 }  // namespace paddle
diff --git a/paddle/function/CMakeLists.txt b/paddle/function/CMakeLists.txt
index 37c011549eca9c8cb9e248051424ad2418770348..31c395c8484a35f9db06dd91626302856414958d 100644
--- a/paddle/function/CMakeLists.txt
+++ b/paddle/function/CMakeLists.txt
@@ -21,6 +21,7 @@ if(WITH_TESTING)
     add_simple_unittest(TensorShapeTest)
     add_simple_unittest(TensorTypeTest)
     add_simple_unittest(BufferArgTest)
+    add_simple_unittest(FunctionTest)
     # add_unittest(ContextProjectionOpTest
     #    ContextProjectionOpTest.cpp
     #    ../gserver/tests/TestUtil.cpp)
diff --git a/paddle/function/Function.cpp b/paddle/function/Function.cpp
index 6f82a8d053bc203eed44bd0d8d4c47d23a15268d..2f56cfc1b5492c23d596f4bfb5a7ae9f066bd10b 100644
--- a/paddle/function/Function.cpp
+++ b/paddle/function/Function.cpp
@@ -72,6 +72,18 @@ FuncConfig& FuncConfig::set<bool>(const std::string& key, bool v) {
   return *this;
 }
 
+void BufferArgs::addArg(const Matrix& arg, const TensorShape& shape) {
+  args_.push_back(std::make_shared<BufferArg>(arg, shape));
+}
+
+void BufferArgs::addArg(const CpuSparseMatrix& arg) {
+  args_.push_back(std::make_shared<SparseMatrixArg>(arg));
+}
+
+void BufferArgs::addArg(const GpuSparseMatrix& arg) {
+  args_.push_back(std::make_shared<SparseMatrixArg>(arg));
+}
+
 ClassRegistrar<FunctionBase> FunctionBase::funcRegistrar_;
 
 }  // namespace paddle
diff --git a/paddle/function/FunctionTest.cpp b/paddle/function/FunctionTest.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..7c3d6684cded1db1801cb18d637bbd4a976298ca
--- /dev/null
+++ b/paddle/function/FunctionTest.cpp
@@ -0,0 +1,59 @@
+/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License. */
+
+#include "Function.h"
+#include <gtest/gtest.h>
+
+namespace paddle {
+
+template <DeviceType DType>
+void FunctionApi(typename Tensor<real, DType>::Matrix& output,
+                 const typename Tensor<real, DType>::Matrix& input);
+
+template <>
+void FunctionApi<DEVICE_TYPE_CPU>(CpuMatrix& output, const CpuMatrix& input) {
+  EXPECT_EQ(output.getHeight(), 100);
+  EXPECT_EQ(output.getWidth(), 200);
+}
+
+template <>
+void FunctionApi<DEVICE_TYPE_GPU>(GpuMatrix& output, const GpuMatrix& input) {
+  EXPECT_EQ(output.getHeight(), 10);
+  EXPECT_EQ(output.getWidth(), 20);
+}
+
+template <DeviceType DType>
+void Function(const BufferArgs& arguments) {
+  auto input = arguments[0].matrix<DType>();
+  auto output = arguments[1].matrix<DType>();
+  FunctionApi<DType>(output, input);
+}
+
+TEST(Function, BufferArgs) {
+  CpuMatrix cpuInput = CpuMatrix(100, 200);
+  CpuMatrix cpuOutput = CpuMatrix(100, 200);
+  BufferArgs cpuArgments;
+  cpuArgments.addArg(cpuInput);
+  cpuArgments.addArg(cpuOutput);
+  Function<DEVICE_TYPE_CPU>(cpuArgments);
+
+  GpuMatrix gpuInput = GpuMatrix(10, 20);
+  GpuMatrix gpuOutput = GpuMatrix(10, 20);
+  BufferArgs gpuArgments;
+  gpuArgments.addArg(gpuInput);
+  gpuArgments.addArg(gpuOutput);
+  Function<DEVICE_TYPE_GPU>(gpuArgments);
+}
+
+}  // namespace paddle