diff --git a/test/npu/test_multi_devices_single_process.py b/test/distributed/test_multi_devices_single_process.py
similarity index 95%
rename from test/npu/test_multi_devices_single_process.py
rename to test/distributed/test_multi_devices_single_process.py
index 4b987aa4f836bd627a98745b1652f052b3a65cf6..63afacc1d7d38b9ddca32378426e9b09aaf60cce 100644
--- a/test/npu/test_multi_devices_single_process.py
+++ b/test/distributed/test_multi_devices_single_process.py
@@ -175,6 +175,13 @@ class TestOp(TestCase):
         custom_output = self._custom_op_exec(query, key, value, head_dim)
         self.assertRtolEqual(supported_output, custom_output)
 
+    def _test_ones_like(self, device="npu:1"):
+        torch.npu.set_device(0)
+        expected = torch.ones(100, 100, device=device)
+        x = torch.randn(100, 100, device=device, dtype=torch.float32)
+        output = torch.ones_like(x)
+        self.assertEqual(output, expected)
+
     @skipIfUnsupportMultiNPU(2)
     def test_aclop_op_with_multi_device(self):
         torch.npu.set_compile_mode(jit_compile=True)
@@ -182,6 +189,7 @@ class TestOp(TestCase):
         self._test_isfinite()
         self._test_unique_dim()
         self._test_npu_prompt_flash_attention()
+        self._test_ones_like()
 
     @skipIfUnsupportMultiNPU(2)
     def test_opapi_op_with_multi_device(self):
@@ -190,6 +198,7 @@ class TestOp(TestCase):
         self._test_isfinite()
         self._test_unique_dim()
         self._test_npu_prompt_flash_attention()
+        self._test_ones_like()
 
 
 if __name__ == '__main__':