diff --git a/python/paddle/fluid/tests/unittests/op_test.py b/python/paddle/fluid/tests/unittests/op_test.py
index 0200d74136ffb0921ec651fe5a6f8f566d36b552..76a707efdc0804be0316ab12c347ffed6199529a 100644
--- a/python/paddle/fluid/tests/unittests/op_test.py
+++ b/python/paddle/fluid/tests/unittests/op_test.py
@@ -437,8 +437,17 @@ class OpTest(unittest.TestCase):
         op_inputs = self.inputs if hasattr(self, "inputs") else dict()
         op_outputs = self.outputs if hasattr(self, "outputs") else dict()
         op_attrs = self.attrs if hasattr(self, "attrs") else dict()
-        self.op = create_op(self.scope, self.op_type, op_inputs, op_outputs,
-                            op_attrs)
+
+        cache_list = None
+        if hasattr(self, "cache_name_list"):
+            cache_list = self.cache_name_list
+        self.op = create_op(
+            self.scope,
+            self.op_type,
+            op_inputs,
+            op_outputs,
+            op_attrs,
+            cache_list=cache_list)
 
         if no_grad_set is None:
             no_grad_set = set()
diff --git a/python/paddle/fluid/tests/unittests/test_lstm_cudnn_op.py b/python/paddle/fluid/tests/unittests/test_lstm_cudnn_op.py
index 9c8e7beae2733922b8520353c48155242ae08a28..0e9e2e8429e51a328e397f9e2a05ab7209c9c1a2 100644
--- a/python/paddle/fluid/tests/unittests/test_lstm_cudnn_op.py
+++ b/python/paddle/fluid/tests/unittests/test_lstm_cudnn_op.py
@@ -121,9 +121,9 @@ class TestCUDNNLstmOp(OpTest):
         self.op_type = "cudnn_lstm"
         self.dtype = np.float32
 
-        num_steps = 50
-        batch_size = 20
-        hidden_size = 200
+        num_steps = 20
+        batch_size = 5
+        hidden_size = 20
 
         input_weight_size = (hidden_size * hidden_size) * 4
         hidden_weight_size = (hidden_size * hidden_size) * 4
@@ -175,6 +175,15 @@ class TestCUDNNLstmOp(OpTest):
             place = core.CUDAPlace(0)
             self.check_output_with_place(place, atol=1e-5)
 
+    def test_grad_with_place(self):
+        if core.is_compiled_with_cuda():
+            place = core.CUDAPlace(0)
+            self.check_grad_with_place(
+                place,
+                set(['Input', 'W', 'InitH', 'InitC']),
+                ['Out', 'last_h', 'last_c'],
+                max_relative_error=0.02)
+
     def testcuda(self):
         return core.is_compiled_with_cuda()
 
diff --git a/python/paddle/fluid/tests/unittests/testsuite.py b/python/paddle/fluid/tests/unittests/testsuite.py
index 34fbb1b549cf5fc5f75bcc0715e5c83665f1d200..dc3b2cb8bc15836a4bf067caa05c3a37a917ecad 100644
--- a/python/paddle/fluid/tests/unittests/testsuite.py
+++ b/python/paddle/fluid/tests/unittests/testsuite.py
@@ -20,7 +20,7 @@ import paddle.fluid.core as core
 from paddle.fluid.op import Operator
 
 
-def create_op(scope, op_type, inputs, outputs, attrs):
+def create_op(scope, op_type, inputs, outputs, attrs, cache_list=None):
     kwargs = dict()
 
     op_maker = core.op_proto_and_checker_maker
@@ -43,6 +43,11 @@ def create_op(scope, op_type, inputs, outputs, attrs):
                     __create_var__(in_name, sub_in_name)
             else:
                 __create_var__(in_name, in_name)
+    if cache_list != None and isinstance(cache_list, list):
+        for name in cache_list:
+            kwargs[name] = []
+            scope.var(name)
+            kwargs[name].append(name)
 
     for out_name, out_dup in Operator.get_op_outputs(op_type):
         if out_name in outputs: