[CI] Add pytest markers to current tests and update the doc. (#577)

Signed-off-by: Alicia <115451386+congw729@users.noreply.github.com>
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com>

tests/e2e/offline_inference/test_cache_dit.py

@@ -15,6 +15,7 @@ from pathlib import Path
 import pytest
 import torch
 
+from tests.utils import hardware_test
 from vllm_omni.inputs.data import OmniDiffusionSamplingParams
 
 # ruff: noqa: E402
@@ -32,6 +33,10 @@ os.environ["VLLM_TEST_CLEAN_GPU_MEMORY"] = "1"
 models = ["riverclouds/qwen_image_random"]
 
 
+@pytest.mark.core_model
+@pytest.mark.diffusion
+@pytest.mark.cache
+@hardware_test(res={"cuda": "L4", "rocm": "MI325"})
 @pytest.mark.parametrize("model_name", models)
 def test_cache_dit(model_name: str):
     """Test cache-dit backend with diffusion model."""

tests/e2e/offline_inference/test_diffusion_cpu_offload.py

@@ -5,7 +5,7 @@ import pytest
 import torch
 from vllm.distributed.parallel_state import cleanup_dist_env_and_memory
 
-from tests.utils import GPUMemoryMonitor
+from tests.utils import GPUMemoryMonitor, hardware_test
 from vllm_omni.inputs.data import OmniDiffusionSamplingParams
 from vllm_omni.platforms import current_omni_platform
 
@@ -45,6 +45,9 @@ def inference(model_name: str, offload: bool = True):
     return peak
 
 
+@pytest.mark.core_model
+@pytest.mark.diffusion
+@hardware_test(res={"cuda": "L4", "rocm": "MI325"})
 @pytest.mark.skipif(current_omni_platform.is_npu() or current_omni_platform.is_rocm(), reason="Hardware not supported")
 @pytest.mark.parametrize("model_name", models)
 def test_cpu_offload_diffusion_model(model_name: str):

tests/e2e/offline_inference/test_qwen2_5_omni.py

@@ -13,10 +13,10 @@ from vllm.assets.video import VideoAsset
 from vllm.envs import VLLM_USE_MODELSCOPE
 from vllm.multimodal.image import convert_image_mode
 
+from tests.utils import create_new_process_for_each_test, hardware_test
 from vllm_omni.platforms import current_omni_platform
 
 from .conftest import OmniRunner
-from .utils import create_new_process_for_each_test
 
 models = ["Qwen/Qwen2.5-Omni-3B"]
 
@@ -34,8 +34,10 @@ test_params = [(model, stage_config) for model in models]
 
 
 @pytest.mark.core_model
-@pytest.mark.parametrize("test_config", test_params)
+@pytest.mark.omni
+@hardware_test(res={"cuda": "L4", "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
 @create_new_process_for_each_test("spawn")
+@pytest.mark.parametrize("test_config", test_params)
 def test_mixed_modalities_to_audio(omni_runner: type[OmniRunner], test_config: tuple[str, str]) -> None:
     """Test processing audio, image, and video together, generating audio output."""
     model, stage_config_path = test_config
@@ -94,8 +96,10 @@ def test_mixed_modalities_to_audio(omni_runner: type[OmniRunner], test_config: t
 
 
 @pytest.mark.core_model
-@pytest.mark.parametrize("test_config", test_params)
+@pytest.mark.omni
+@hardware_test(res={"cuda": "L4", "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
 @create_new_process_for_each_test("spawn")
+@pytest.mark.parametrize("test_config", test_params)
 def test_mixed_modalities_to_text_only(omni_runner: type[OmniRunner], test_config: tuple[str, str]) -> None:
     """Test processing audio, image, and video together, generating audio output."""
     model, stage_config_path = test_config

tests/e2e/offline_inference/test_qwen3_omni.py

@@ -14,6 +14,7 @@ from pathlib import Path
 import pytest
 from vllm.assets.video import VideoAsset
 
+from tests.utils import hardware_test
 from vllm_omni.platforms import current_omni_platform
 
 from .conftest import OmniRunner
@@ -31,6 +32,9 @@ else:
 test_params = [(model, stage_config) for model in models for stage_config in stage_configs]
 
 
+@pytest.mark.core_model
+@pytest.mark.omni
+@hardware_test(res={"cuda": "H100", "rocm": "MI325"}, num_cards=2)
 @pytest.mark.parametrize("test_config", test_params)
 def test_video_to_audio(omni_runner: type[OmniRunner], test_config) -> None:
     """Test processing video, generating audio output."""

tests/e2e/offline_inference/test_sequence_parallel.py

@@ -20,16 +20,17 @@ import torch
 import torch.distributed as dist
 from PIL import Image
 
+from tests.utils import hardware_test
+from vllm_omni import Omni
+from vllm_omni.diffusion.data import DiffusionParallelConfig
 from vllm_omni.inputs.data import OmniDiffusionSamplingParams
+from vllm_omni.platforms import current_omni_platform
 
 # ruff: noqa: E402
 REPO_ROOT = Path(__file__).resolve().parents[3]
 if str(REPO_ROOT) not in sys.path:
     sys.path.insert(0, str(REPO_ROOT))
 
-from vllm_omni import Omni
-from vllm_omni.diffusion.data import DiffusionParallelConfig
-from vllm_omni.platforms import current_omni_platform
 
 # Test configuration
 MODELS = ["riverclouds/qwen_image_random"]
@@ -145,9 +146,11 @@ def _run_inference(
 # - warmup: whether to run warmup for this SP config
 # - is_perf_test: whether this is a performance test (show speedup metrics)
 SP_CONFIGS = [
-    (2, 1, DEFAULT_HEIGHT, DEFAULT_WIDTH, True, True),  # Ulysses-2 - performance test
+    # Ulysses-2 - performance test
+    (2, 1, DEFAULT_HEIGHT, DEFAULT_WIDTH, True, True),
     (1, 2, DEFAULT_HEIGHT, DEFAULT_WIDTH, True, True),  # Ring-2 - performance test
-    (2, 2, DEFAULT_HEIGHT, DEFAULT_WIDTH, False, False),  # Hybrid - correctness only
+    # Hybrid - correctness only
+    (2, 2, DEFAULT_HEIGHT, DEFAULT_WIDTH, False, False),
     (4, 1, 272, 272, False, False),  # Ulysses-4 - shape and correctness
 ]
 
@@ -162,6 +165,10 @@ def _get_sp_mode(ulysses_degree: int, ring_degree: int) -> str:
         return f"hybrid-{ulysses_degree}x{ring_degree}"
 
 
+@pytest.mark.core_model
+@pytest.mark.diffusion
+@pytest.mark.parallel
+@hardware_test(res={"cuda": "L4", "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
 @pytest.mark.parametrize("model_name", MODELS)
 def test_sp_correctness(model_name: str):
     """Test that SP inference produces correct outputs and measure performance.

tests/e2e/offline_inference/test_stable_audio_model.py

@@ -5,6 +5,7 @@ import numpy as np
 import pytest
 import torch
 
+from tests.utils import hardware_test
 from vllm_omni.inputs.data import OmniDiffusionSamplingParams
 from vllm_omni.outputs import OmniRequestOutput
 
@@ -19,6 +20,9 @@ from vllm_omni import Omni
 models = ["linyueqian/stable_audio_random"]
 
 
+@pytest.mark.core_model
+@pytest.mark.diffusion
+@hardware_test(res={"cuda": "L4"})
 @pytest.mark.parametrize("model_name", models)
 def test_stable_audio_model(model_name: str):
     m = Omni(model=model_name)

tests/e2e/offline_inference/test_t2i_model.py

@@ -5,6 +5,7 @@ from pathlib import Path
 import pytest
 import torch
 
+from tests.utils import hardware_test
 from vllm_omni.inputs.data import OmniDiffusionSamplingParams
 from vllm_omni.outputs import OmniRequestOutput
 from vllm_omni.platforms import current_omni_platform
@@ -32,6 +33,9 @@ elif current_omni_platform.is_rocm():
     models = ["Tongyi-MAI/Z-Image-Turbo"]
 
 
+@pytest.mark.core_model
+@pytest.mark.diffusion
+@hardware_test(res={"cuda": "L4", "rocm": "MI325"}, num_cards={"cuda": 1, "rocm": 2})
 @pytest.mark.parametrize("model_name", models)
 def test_diffusion_model(model_name: str):
     m = None

tests/e2e/offline_inference/test_teacache.py

@@ -15,6 +15,7 @@ from pathlib import Path
 import pytest
 import torch
 
+from tests.utils import hardware_test
 from vllm_omni.inputs.data import OmniDiffusionSamplingParams
 
 # ruff: noqa: E402
@@ -31,6 +32,10 @@ os.environ["VLLM_TEST_CLEAN_GPU_MEMORY"] = "1"
 models = ["riverclouds/qwen_image_random"]
 
 
+@pytest.mark.core_model
+@pytest.mark.diffusion
+@pytest.mark.cache
+@hardware_test(res={"cuda": "L4", "rocm": "MI325"})
 @pytest.mark.parametrize("model_name", models)
 def test_teacache(model_name: str):
     """Test TeaCache backend with diffusion model."""

tests/e2e/offline_inference/test_zimage_tensor_parallel.py

@@ -12,18 +12,18 @@ import torch
 from PIL import Image
 from vllm.distributed.parallel_state import cleanup_dist_env_and_memory
 
+from tests.utils import GPUMemoryMonitor, hardware_test
+from vllm_omni import Omni
+from vllm_omni.diffusion.data import DiffusionParallelConfig
 from vllm_omni.inputs.data import OmniDiffusionSamplingParams
+from vllm_omni.outputs import OmniRequestOutput
+from vllm_omni.platforms import current_omni_platform
 
 # ruff: noqa: E402
 REPO_ROOT = Path(__file__).resolve().parents[2]
 if str(REPO_ROOT) not in sys.path:
     sys.path.insert(0, str(REPO_ROOT))
 
-from tests.utils import GPUMemoryMonitor
-from vllm_omni import Omni
-from vllm_omni.diffusion.data import DiffusionParallelConfig
-from vllm_omni.outputs import OmniRequestOutput
-from vllm_omni.platforms import current_omni_platform
 
 # os.environ["VLLM_TEST_CLEAN_GPU_MEMORY"] = "1"
 os.environ["VLLM_WORKER_MULTIPROC_METHOD"] = "spawn"
@@ -127,7 +127,10 @@ def _run_zimage_generate(
         cleanup_dist_env_and_memory()
 
 
-@pytest.mark.integration
+@pytest.mark.core_model
+@pytest.mark.diffusion
+@pytest.mark.parallel
+@hardware_test(res={"cuda": "L4", "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
 def test_zimage_tensor_parallel_tp2(tmp_path: Path):
     if current_omni_platform.is_npu() or current_omni_platform.is_rocm():
         pytest.skip("Z-Image TP e2e test is only supported on CUDA for now.")

tests/e2e/online_serving/test_async_omni.py

@@ -8,6 +8,7 @@ import pytest
 from vllm import SamplingParams
 from vllm.inputs import PromptType
 
+from tests.utils import hardware_test
 from vllm_omni.entrypoints.async_omni import AsyncOmni, ClientRequestState
 
 os.environ["VLLM_WORKER_MULTIPROC_METHOD"] = "spawn"
@@ -57,6 +58,9 @@ async def generate(
     return count, request_id
 
 
+@pytest.mark.core_model
+@pytest.mark.omni
+@hardware_test(res={"cuda": "H100", "rocm": "MI325"}, num_cards=2)
 @pytest.mark.asyncio
 async def test_abort():
     with ExitStack() as after:
@@ -113,6 +117,9 @@ async def test_abort():
     await asyncio.sleep(5)
 
 
+@pytest.mark.core_model
+@pytest.mark.omni
+@hardware_test(res={"cuda": "H100", "rocm": "MI325"}, num_cards=2)
 @pytest.mark.asyncio
 async def test_build_and_log_summary(monkeypatch):
     from vllm_omni.entrypoints.utils import get_final_stage_id_for_e2e

tests/e2e/online_serving/test_image_gen_edit.py

@@ -22,6 +22,8 @@ from PIL import Image
 from vllm.assets.image import ImageAsset
 from vllm.utils.network_utils import get_open_port
 
+from tests.utils import hardware_test
+
 os.environ["VLLM_WORKER_MULTIPROC_METHOD"] = "spawn"
 # Increase timeout for downloading assets from S3 (default 5s is too short for CI)
 os.environ.setdefault("VLLM_IMAGE_FETCH_TIMEOUT", "60")
@@ -178,6 +180,9 @@ def _decode_data_url_to_image_bytes(data_url: str) -> bytes:
     return base64.b64decode(b64_data)
 
 
+@pytest.mark.core_model
+@pytest.mark.diffusion
+@hardware_test(res={"cuda": "H100", "rocm": "MI325"})
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_i2i_multi_image_input_qwen_image_edit_2509(
     omni_server,

tests/e2e/online_serving/test_images_generations_lora.py

@@ -23,6 +23,7 @@ from PIL import Image
 from safetensors.torch import save_file
 
 from tests.conftest import OmniServer
+from tests.utils import hardware_test
 
 os.environ["VLLM_WORKER_MULTIPROC_METHOD"] = "spawn"
 
@@ -144,6 +145,9 @@ def _basic_payload() -> dict:
     }
 
 
+@pytest.mark.core_model
+@pytest.mark.diffusion
+@hardware_test(res={"cuda": "L4", "rocm": "MI325"})
 def test_images_generations_per_request_lora_switching(omni_server: OmniServer, tmp_path: Path) -> None:
     # Base generation.
     base_img = _post_images(omni_server, _basic_payload())

tests/e2e/online_serving/test_qwen3_omni.py

@@ -28,6 +28,7 @@ from tests.conftest import (
     merge_base64_and_convert_to_text,
     modify_stage_config,
 )
+from tests.utils import hardware_test
 from vllm_omni.platforms import current_omni_platform
 
 models = ["Qwen/Qwen3-Omni-30B-A3B-Instruct"]
@@ -145,6 +146,9 @@ def get_max_batch_size(size_type="few"):
     return batch_sizes.get(size_type, 5)
 
 
+@pytest.mark.core_model
+@pytest.mark.omni
+@hardware_test(res={"cuda": "H100", "rocm": "MI325"}, num_cards=2)
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_mix_to_text_audio_001(client: openai.OpenAI, omni_server, request) -> None:
     """
@@ -215,6 +219,9 @@ def test_mix_to_text_audio_001(client: openai.OpenAI, omni_server, request) -> N
     assert similarity > 0.9, "The audio content is not same as the text"
 
 
+@pytest.mark.core_model
+@pytest.mark.omni
+@hardware_test(res={"cuda": "H100", "rocm": "MI325"}, num_cards=2)
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_text_to_text_audio_001(client: openai.OpenAI, omni_server) -> None:
     """