[TEST] add xlite e2e test

Signed-off-by: lulina <[email protected]>

Author: lulina

.github/workflows/_e2e_test.yaml

@@ -103,6 +103,7 @@ jobs:
           pytest -sv tests/e2e/singlecard/test_sampler.py
           pytest -sv tests/e2e/singlecard/test_vlm.py
           pytest -sv tests/e2e/singlecard/multi-modal/test_internvl.py
+          pytest -sv tests/e2e/singlecard/test_xlite.py
 
           # ------------------------------------ v1 spec decode test ------------------------------------ #
           pytest -sv tests/e2e/singlecard/spec_decode_v1/test_v1_mtp_correctness.py

requirements-dev.txt

@@ -20,4 +20,5 @@ soundfile
 pytest_mock
 msserviceprofiler>=1.2.2
 mindstudio-probe>=8.3.0
-arctic-inference==0.1.1
+arctic-inference==0.1.1
+xlite

tests/e2e/singlecard/test_xlite.py

@@ -0,0 +1,130 @@
+#
+# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
+# Copyright 2023 The vLLM team.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+"""
+Compare the outputs of vLLM with and without xlite.
+
+Run `pytest tests/e2e/singlecard/test_xlite.py`.
+"""
+
+import pytest
+from vllm import SamplingParams
+
+from tests.e2e.conftest import VllmRunner
+from tests.e2e.model_utils import check_outputs_equal
+
+MODELS = [
+    "Qwen/Qwen3-0.6B",
+]
+
+
+@pytest.mark.parametrize("model", MODELS)
+@pytest.mark.parametrize("max_tokens", [32])
+def test_models_with_xlite_decode_only(
+    model: str,
+    max_tokens: int,
+) -> None:
+    prompts = [
+        "Hello, my name is", "The president of the United States is",
+        "The capital of France is", "The future of AI is"
+    ]
+
+    sampling_params = SamplingParams(max_tokens=max_tokens, temperature=0.0)
+    with VllmRunner(
+            model,
+            block_size=128,
+            max_model_len=1024,
+            enforce_eager=False,
+            additional_config={"xlite_graph_config": {
+                "enabled": True
+            }},
+    ) as runner:
+        vllm_xlite_outputs = runner.model.generate(prompts, sampling_params)
+
+    with VllmRunner(
+            model,
+            block_size=128,
+            max_model_len=1024,
+            enforce_eager=True,
+    ) as runner:
+        vllm_eager_outputs = runner.model.generate(prompts, sampling_params)
+    vllm_xlite_outputs_list = []
+    for output in vllm_xlite_outputs:
+        vllm_xlite_outputs_list.append(
+            (output.outputs[0].index, output.outputs[0].text))
+
+    vllm_eager_outputs_list = []
+    for output in vllm_eager_outputs:
+        vllm_eager_outputs_list.append(
+            (output.outputs[0].index, output.outputs[0].text))
+
+    check_outputs_equal(
+        outputs_0_lst=vllm_eager_outputs_list,
+        outputs_1_lst=vllm_xlite_outputs_list,
+        name_0="vllm_eager_outputs",
+        name_1="vllm_xlite_outputs",
+    )
+
+
+@pytest.mark.parametrize("model", MODELS)
+@pytest.mark.parametrize("max_tokens", [32])
+def test_models_with_xlite_full_mode(
+    model: str,
+    max_tokens: int,
+) -> None:
+    prompts = [
+        "Hello, my name is", "The president of the United States is",
+        "The capital of France is", "The future of AI is"
+    ]
+
+    sampling_params = SamplingParams(max_tokens=max_tokens, temperature=0.0)
+    with VllmRunner(
+            model,
+            block_size=128,
+            max_model_len=1024,
+            enforce_eager=False,
+            additional_config={
+                "xlite_graph_config": {
+                    "enabled": True,
+                    "full_mode": True
+                }
+            },
+    ) as runner:
+        vllm_xlite_outputs = runner.model.generate(prompts, sampling_params)
+
+    with VllmRunner(
+            model,
+            block_size=128,
+            max_model_len=1024,
+            enforce_eager=True,
+    ) as runner:
+        vllm_eager_outputs = runner.model.generate(prompts, sampling_params)
+    vllm_xlite_outputs_list = []
+    for output in vllm_xlite_outputs:
+        vllm_xlite_outputs_list.append(
+            (output.outputs[0].index, output.outputs[0].text))
+
+    vllm_eager_outputs_list = []
+    for output in vllm_eager_outputs:
+        vllm_eager_outputs_list.append(
+            (output.outputs[0].index, output.outputs[0].text))
+
+    check_outputs_equal(
+        outputs_0_lst=vllm_eager_outputs_list,
+        outputs_1_lst=vllm_xlite_outputs_list,
+        name_0="vllm_eager_outputs",
+        name_1="vllm_xlite_outputs",
+    )

tests/ut/test_platform.py

@@ -32,6 +32,7 @@ def mock_vllm_config():
     def mock_vllm_ascend_config():
         mock_ascend_config = MagicMock()
         mock_ascend_config.torchair_graph_config.enabled = False
+        mock_ascend_config.xlite_graph_config.enabled = False
         mock_ascend_config.enable_shared_expert_dp = False
         return mock_ascend_config
 
@@ -512,6 +513,16 @@ def test_check_and_update_config_v1_worker_class_selection(
             "vllm_ascend.torchair.torchair_worker.NPUTorchairWorker",
         )
 
+        test_ascend_config = TestNPUPlatform.mock_vllm_ascend_config()
+        test_ascend_config.xlite_graph_config.enabled = True
+        mock_init_ascend.return_value = test_ascend_config
+        vllm_config.parallel_config.worker_cls = "auto"
+        self.platform.check_and_update_config(vllm_config)
+        self.assertEqual(
+            vllm_config.parallel_config.worker_cls,
+            "vllm_ascend.xlite.xlite_worker.XliteWorker",
+        )
+
     @patch("vllm_ascend.ascend_config.check_ascend_config")
     @patch("vllm_ascend.ascend_config.init_ascend_config")
     @patch('vllm_ascend.utils.get_ascend_device_type',