[Test] Remove VLLM_USE_V1 in example and tests (#1733)

V1 is enabled by default, no need to set it by hand now. This PR remove
the useless setting in example and tests

- vLLM version: v0.9.2
- vLLM main:
9ad0a4588b

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>

tests/e2e/multicard/test_fused_moe_allgather_ep.py

@@ -26,12 +26,11 @@ from unittest.mock import patch
 from modelscope import snapshot_download  # type: ignore
 from vllm import SamplingParams
 
-from tests.conftest import VllmRunner
+from tests.e2e.conftest import VllmRunner
 
 
 @patch.dict(
     os.environ, {
-        "VLLM_USE_V1": "1",
         "VLLM_WORKER_MULTIPROC_METHOD": "spawn",
         "TASK_QUEUE_ENABLE": "1",
         "VLLM_ENABLE_FUSED_EXPERTS_ALLGATHER_EP": "1"
@@ -56,12 +55,10 @@ def test_generate_with_allgather():
         vllm_model.generate(example_prompts, sampling_params)
 
 
-@patch.dict(
-    os.environ, {
-        "VLLM_USE_V1": "1",
-        "VLLM_WORKER_MULTIPROC_METHOD": "spawn",
-        "TASK_QUEUE_ENABLE": "1"
-    })
+@patch.dict(os.environ, {
+    "VLLM_WORKER_MULTIPROC_METHOD": "spawn",
+    "TASK_QUEUE_ENABLE": "1"
+})
 def test_generate_with_alltoall():
     example_prompts = ["Hello, my name is"]
     sampling_params = SamplingParams(max_tokens=100, temperature=0.0)
@@ -79,4 +76,4 @@ def test_generate_with_alltoall():
                         },
                         "expert_tensor_parallel_size": 1
                     }) as vllm_model:
-        vllm_model.generate(example_prompts, sampling_params)
+        vllm_model.generate(example_prompts, sampling_params)

tests/e2e/multicard/test_ilama_lora_tp2.py

@@ -1,7 +1,7 @@
 import pytest
 from modelscope import snapshot_download  # type: ignore
 
-from tests.conftest import VllmRunner
+from tests.e2e.conftest import VllmRunner
 from tests.e2e.singlecard.test_ilama_lora import (EXPECTED_LORA_OUTPUT,
                                                   MODEL_PATH, do_sample)
 

tests/e2e/multicard/test_offline_inference_distributed.py

@@ -27,7 +27,7 @@ from modelscope import snapshot_download  # type: ignore
 from vllm import SamplingParams
 from vllm.model_executor.models.registry import ModelRegistry
 
-from tests.conftest import VllmRunner
+from tests.e2e.conftest import VllmRunner
 
 os.environ["PYTORCH_NPU_ALLOC_CONF"] = "max_split_size_mb:256"
 

tests/e2e/multicard/test_pipeline_parallel.py

@@ -16,7 +16,7 @@
 #
 import pytest
 
-from tests.conftest import VllmRunner
+from tests.e2e.conftest import VllmRunner
 
 MODELS = [
     "Qwen/Qwen3-0.6B",

tests/e2e/multicard/test_prefix_caching.py

@@ -2,12 +2,10 @@
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 """Compare the with and without prefix caching on V1 scheduler or AscendScheduler."""
 
-import os
-
 import pytest
 
-from tests.conftest import VllmRunner
-from tests.model_utils import check_outputs_equal
+from tests.e2e.conftest import VllmRunner
+from tests.e2e.model_utils import check_outputs_equal
 
 MODELS = [
     # for MHA
@@ -60,8 +58,6 @@ INPUT_PROMPTS = [
 ]
 
 
-@pytest.mark.skipif(os.getenv("VLLM_USE_V1") == "0",
-                    reason="mtp is not supported on v1")
 @pytest.mark.parametrize("model", MODELS)
 @pytest.mark.parametrize("max_tokens", [50])
 def test_prefix_cache_with_v1_scheduler(model: str, max_tokens: int) -> None:
@@ -89,8 +85,6 @@ def test_prefix_cache_with_v1_scheduler(model: str, max_tokens: int) -> None:
     )
 
 
-@pytest.mark.skipif(os.getenv("VLLM_USE_V1") == "0",
-                    reason="mtp is not supported on v1")
 @pytest.mark.parametrize("model", MODELS)
 @pytest.mark.parametrize("max_tokens", [50])
 def test_prefix_cache_with_ascend_scheduler(model: str,

tests/e2e/multicard/test_torchair_graph_mode.py

@@ -22,9 +22,7 @@ Run `pytest tests/multicard/test_torchair_graph_mode.py`.
 import os
 from typing import Dict
 
-import pytest
-
-from tests.conftest import VllmRunner
+from tests.e2e.conftest import VllmRunner
 
 os.environ["PYTORCH_NPU_ALLOC_CONF"] = "max_split_size_mb:256"
 
@@ -78,8 +76,6 @@ def _deepseek_torchair_test_fixture(
         print(f"Generated text: {vllm_output[i][1]!r}")
 
 
-@pytest.mark.skipif(os.getenv("VLLM_USE_V1") == "0",
-                    reason="torchair graph is not supported on v0")
 def test_e2e_deepseekv3_with_torchair():
     additional_config = {
         "torchair_graph_config": {
@@ -89,8 +85,6 @@ def test_e2e_deepseekv3_with_torchair():
     _deepseek_torchair_test_fixture(additional_config)
 
 
-@pytest.mark.skipif(os.getenv("VLLM_USE_V1") == "0",
-                    reason="torchair graph is not supported on v0")
 def test_e2e_deepseekv3_with_torchair_ms_mla():
     additional_config = {
         "torchair_graph_config": {
@@ -150,8 +144,6 @@ def _pangu_torchair_test_fixture(
         print(f"Generated text: {vllm_output[i][1]!r}")
 
 
-@pytest.mark.skipif(os.getenv("VLLM_USE_V1") == "0",
-                    reason="torchair graph is not supported on v0")
 def test_e2e_pangu_with_torchair():
     additional_config = {
         "torchair_graph_config": {