diff --git a/tests/ut/ge/hybrid/known_node_executor_unittest.cc b/tests/ut/ge/hybrid/known_node_executor_unittest.cc
index 98e985f7..7d32f712 100644
--- a/tests/ut/ge/hybrid/known_node_executor_unittest.cc
+++ b/tests/ut/ge/hybrid/known_node_executor_unittest.cc
@@ -88,4 +88,33 @@ TEST_F(UnknownNodeExecutorTest, TestParseAttrForAllocatingOutputs) {
   ASSERT_EQ(node_item.ref_outputs[1], const_node);
   ASSERT_EQ(node_item.reuse_inputs.size(), 1);
   ASSERT_EQ(node_item.reuse_inputs[0], 0);
+}
+
+TEST_F(UnknownNodeExecutorTest, TestSetGlobalStepInLoadTask) {
+OpDescPtr op_desc = CreateOpDesc("PartitionedCall", "PartitionedCall");
+auto node = root_graph->AddNode(op_desc);
+node->SetOwnerComputeGraph(root_graph);
+auto sub_graph = BuildDataDirectConnectGraph();
+sub_graph->SetParentGraph(root_graph);
+sub_graph->SetParentNode(node);
+node->GetOpDesc()->AddSubgraphName("subgraph");
+node->GetOpDesc()->SetSubgraphInstanceName(0, "subgraph");
+root_graph->AddSubgraph("subgraph", sub_graph);
+
+ComputeGraphPtr root_graph = std::make_shared<ComputeGraph>("root_graph");
+GeRootModelPtr ge_root_model = make_shared<GeRootModel>(root_graph);
+HybridModel hybrid_model(ge_root_model);
+hybrid_model.model_id_ = 0;
+hybrid_model.model_name_ = "root_model";
+hybrid_model.om_name_ = "temp_om";
+auto *step_id = new int64_t[1];
+step_id[0] = 520;
+TensorValue tensor_value((void *)step_id, sizeof(step_id));
+hybrid_model.variable_tensors_.insert({"ge_global_step", make_unique<TensorValue>(tensor_value)});
+
+KnownNodeExecutor known_node_executor;
+shared_ptr<KnownNodeTask> task = nullptr;
+Status ret = known_node_executor.LoadTask(hybrid_model, node, task);
+EXPECT_EQ(*(task->davinci_model_->global_step_addr_), 520);
+EXPECT_EQ(known_node_executor.LoadTask(hybrid_model, node, task), SUCCESS);
 }
\ No newline at end of file