diff --git a/mindspore/python/mindspore/ops/operations/array_ops.py b/mindspore/python/mindspore/ops/operations/array_ops.py
index dd7e8f8d3c8..4aec0dd86db 100755
--- a/mindspore/python/mindspore/ops/operations/array_ops.py
+++ b/mindspore/python/mindspore/ops/operations/array_ops.py
@@ -2079,7 +2079,8 @@ class Tile(PrimitiveWithInfer):
             raise TypeError(f"For '{self.name}', the type of 'input_x' should be Tensor, "
                             f"but got {type(base_tensor).__name__}.")
         if all(v == 1 for v in multiplier) and len(base_tensor.shape) >= len(multiplier):
-            return (True, base_tensor.copy())
+            ret = Identity()(base_tensor)
+            return (True, ret)
         return (False, None)
 
     def _get_shape_and_range(self, x, multiples):
diff --git a/tests/st/pynative/test_ops.py b/tests/st/pynative/test_ops.py
index c43e626be5e..cc47d8016cb 100644
--- a/tests/st/pynative/test_ops.py
+++ b/tests/st/pynative/test_ops.py
@@ -1,4 +1,4 @@
-# Copyright 2020 Huawei Technologies Co., Ltd
+# Copyright 2020-2022 Huawei Technologies Co., Ltd
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -12,20 +12,52 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 # ============================================================================
-
+import pytest
 import numpy as np
 
 import mindspore as ms
-import mindspore.ops.operations as P
-from mindspore import context, Tensor
+from mindspore import context
+from mindspore import ops, Tensor, dtype, ms_function
 
 
 def test_cast():
-    """ tests cast for same dtype"""
+    """
+    Feature: test cast operator
+    Description: Cast original data type to target data type
+    Expectation: success
+    """
     context.set_context(mode=context.PYNATIVE_MODE, device_target="Ascend")
     input_np = np.random.randn(2, 3, 4, 5).astype(np.float32)
     input_x = Tensor(input_np)
     type_dst = ms.float32
-    cast = P.Cast()
+    cast = ops.Cast()
     result = cast(input_x, type_dst)
     assert result.dtype == type_dst
+
+
+@ms_function
+def expand_tensor(a, b):
+    out = ops.tile(a, b)
+    return out
+
+
+@pytest.mark.level0
+@pytest.mark.platform_x86_cpu
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.env_onecard
+def test_tile_eliminate():
+    """
+    Feature: tile_eliminate
+    Description: All value of multiplier is '1' but length of multiplier is greater than tensor dims, can't do eliminate
+    Expectation: success
+    """
+    context.set_context(mode=context.PYNATIVE_MODE)
+    tensor_ = Tensor(np.ndarray([1, 448, 448]), dtype=dtype.float32)
+    out = ops.tile(tensor_, (1, 1, 1))
+    assert out.shape == (1, 448, 448)
+    out = ops.tile(tensor_, (1, 1, 1, 1))
+    assert out.shape == (1, 1, 448, 448)
+    out = expand_tensor(tensor_, (1, 1, 1))
+    assert out.shape == (1, 448, 448)
+    out = expand_tensor(tensor_, (1, 1, 1, 1))
+    assert out.shape == (1, 1, 448, 448)
diff --git a/tests/ut/python/optimizer/test_tile_eliminate.py b/tests/ut/python/optimizer/test_tile_eliminate.py
deleted file mode 100644
index c4a29e13bb4..00000000000
--- a/tests/ut/python/optimizer/test_tile_eliminate.py
+++ /dev/null
@@ -1,27 +0,0 @@
-import numpy as np
-from mindspore import context
-from mindspore import ms_function, ops, Tensor, dtype
-
-
-@ms_function
-def expand_tensor(a, b):
-    out = ops.tile(a, b)
-    return out
-
-
-def test_tile_eliminate():
-    """
-    Feature: tile_eliminate
-    Description: All value of multiplier is '1' but length of multiplier is greater than tensor dims, can't do eliminate
-    Expectation: success
-    """
-    context.set_context(mode=context.PYNATIVE_MODE)
-    tensor_ = Tensor(np.ndarray([1, 448, 448]), dtype=dtype.float32)
-    out = ops.tile(tensor_, (1, 1, 1))
-    assert out.shape == (1, 448, 448)
-    out = ops.tile(tensor_, (1, 1, 1, 1))
-    assert out.shape == (1, 1, 448, 448)
-    out = expand_tensor(tensor_, (1, 1, 1))
-    assert out.shape == (1, 448, 448)
-    out = expand_tensor(tensor_, (1, 1, 1, 1))
-    assert out.shape == (1, 1, 448, 448)