diff --git a/test/_inductor/__init__.py b/test/_inductor/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..ac0a39a91feeaed1fe9b67cc972855a8927b99c0
--- /dev/null
+++ b/test/_inductor/__init__.py
@@ -0,0 +1,3 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+#  Copyright (c) Huawei Technologies Co., Ltd. 2025-2025. All rights reserved.
\ No newline at end of file
diff --git a/test/_inductor/test_add.py b/test/_inductor/test_add.py
new file mode 100644
index 0000000000000000000000000000000000000000..f952b20031df6d1d0d8463050a02626be4be5c02
--- /dev/null
+++ b/test/_inductor/test_add.py
@@ -0,0 +1,30 @@
+import torch
+from torch.testing._internal.common_utils import run_tests, parametrize, instantiate_parametrized_tests
+from testutils import TestUtils
+import torch_npu
+import torch_npu._inductor
+
+
+class TestAdd(TestUtils):
+    def op_calc(self, first_element, second_element):
+        result = first_element + second_element
+        return result
+
+    @parametrize('shape', TestUtils._pointwise_demo_shapes)
+    @parametrize('dtype', ['float32', 'int64'])
+    def test_pointwise_cases(self, shape, dtype):
+        first_element = self._generate_tensor(shape, dtype)
+        second_element = self._generate_tensor(shape, dtype)
+
+        std_sum = self.op_calc(first_element, second_element)
+
+        compiled_op_calc = torch.compile(self.op_calc, backend="inductor")
+        inductor_sum = compiled_op_calc(first_element, second_element)
+
+        self.assertEqual(std_sum, inductor_sum)
+
+
+instantiate_parametrized_tests(TestAdd)
+
+if __name__ == "__main__":
+    run_tests()
diff --git a/test/_inductor/testutils.py b/test/_inductor/testutils.py
new file mode 100644
index 0000000000000000000000000000000000000000..47863aaf3483ea0923564d3dff71e3363b31647f
--- /dev/null
+++ b/test/_inductor/testutils.py
@@ -0,0 +1,36 @@
+from collections.abc import Sequence
+import os
+import time
+import numpy as np
+import torch
+from torch.testing._internal.common_utils import TestCase
+import torch_npu
+
+
+class TestUtils(TestCase):
+    _pointwise_test_shape2d = [(4096, 256), (1024, 32), (8, 2048), (8, 4096)]  # (8, 4), (8, 8), not supported
+    _pointwise_test_shape3d = [(8, 8, 4), (8, 8, 8), (8, 8, 2048), (8, 8, 4096)]
+    _pointwise_test_shape4d = [(128, 128, 4096, 4), (128, 128, 4096, 8),
+                               (32, 32, 1024, 1024)]  # 128*128*4096*2048 is too big(512G)
+    _pointwise_test_shapes = _pointwise_test_shape2d + _pointwise_test_shape3d + _pointwise_test_shape4d
+
+    _pointwise_demo_shapes = [(1024, 32), (8, 16, 256, 32)]
+    _reduction_extest_shape4d = [(8, 8, 8, 16384), (8, 8, 16384, 8), (8, 16384, 8, 8), (16384, 8, 8, 8)]
+    _reduction_extest_dim4d = [-1, -2, 1, 0]
+    _reduction_extest_SDbinding = list(zip(_reduction_extest_shape4d, _reduction_extest_dim4d))
+
+    _test_dtypes = ['float32', 'int32', 'float16', 'bfloat16', 'int64']
+
+    @staticmethod
+    def _generate_tensor(shape, dtype, floatPOSIFLAG=0):
+        if dtype == 'float32' or dtype == 'float16' or dtype == 'bfloat16':
+            if floatPOSIFLAG:
+                return 1000 * torch.rand(size=shape, dtype=eval('torch.' + dtype), device=torch.device("npu"))
+            else:
+                return torch.randn(size=shape, dtype=eval('torch.' + dtype), device=torch.device("npu")) * 2000
+        elif dtype == 'int32' or dtype == 'int64':
+            return torch.randint(low=0, high=2000, size=shape, dtype=eval('torch.' + dtype), device=torch.device("npu"))
+        elif dtype == 'bool':
+            return torch.randint(low=0, high=2, size=shape, device=torch.device("npu")).bool()
+        else:
+            raise ValueError('Invalid parameter \"dtype\" is found : {}'.format(dtype))
diff --git a/torch_npu/_inductor/__init__.py b/torch_npu/_inductor/__init__.py
index 04ef05321d0362f1aa3f5396b49fb6311cafaf81..61a50b38f68b3d7a7d428a08cedc61ea4e213d6a 100644
--- a/torch_npu/_inductor/__init__.py
+++ b/torch_npu/_inductor/__init__.py
@@ -78,6 +78,7 @@ if npu_config.dump_fx_graph:
 else:
     from .lowering import _register_npu_inductor_fallbacks
 
+
 _register_npu_inductor_fallbacks()
 _register_npu_inductor_decompositons()