[CI/Build] Share nightly CUDA YAML between H100/L4 and B200 (#7028)

Signed-off-by: wangyu <410167048@qq.com>
Signed-off-by: [Your Name] [Your Email]
Signed-off-by: [Your Name] <your.email@example.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: amy-why-3459 <wuhaiyan17@huawei.com>

tests/examples/offline_inference/test_image_to_video.py

@@ -1,3 +1,6 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
 """
 Offline inference tests: image-to-video.
 See examples/offline_inference/image_to_video/README.md
@@ -17,7 +20,7 @@ pytestmark = [
     pytest.mark.usefixtures("clean_gpu_memory_between_tests"),
     pytest.mark.full_model,
     pytest.mark.example,
-    *hardware_marks(res={"cuda": "H100"}),
+    *hardware_marks(res={"cuda": ["H100", "B200"]}),
 ]
 
 I2V_SCRIPT = EXAMPLES / "offline_inference" / "image_to_video" / "image_to_video.py"

tests/examples/offline_inference/test_internvla_a1.py

@@ -1,5 +1,5 @@
 # SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
 
 """Offline E2E test for InternVLA-A1 open-loop inference."""
 
@@ -26,6 +26,7 @@ def _required_env(name: str) -> str:
     value = os.getenv(name)
     if not value:
         pytest.skip(f"{name} is not set")
+    assert value
     return value
 
 

tests/examples/offline_inference/test_text_to_image.py

@@ -18,7 +18,7 @@ pytestmark = [
     pytest.mark.usefixtures("clean_gpu_memory_between_tests"),
     pytest.mark.full_model,
     pytest.mark.example,
-    *hardware_marks(res={"cuda": "H100"}),
+    *hardware_marks(res={"cuda": ["H100", "B200"]}),
 ]
 
 T2I_SCRIPT = EXAMPLES / "offline_inference" / "text_to_image" / "text_to_image.py"

tests/examples/offline_inference/test_vace_video_generation.py

@@ -1,3 +1,6 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
 """VACE coverage through the shared text-to-video and image-to-video examples."""
 
 from pathlib import Path
@@ -14,7 +17,7 @@ pytestmark = [
     pytest.mark.full_model,
     pytest.mark.diffusion,
     pytest.mark.example,
-    *hardware_marks(res={"cuda": "H100"}),
+    *hardware_marks(res={"cuda": ["H100", "B200"]}),
 ]
 
 TEXT_TO_VIDEO_README = EXAMPLES / "offline_inference" / "text_to_video" / "text_to_video.md"

tests/examples/online_serving/test_qwen2_5_omni.py

@@ -1,3 +1,6 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
 """
 Online serving tests: Qwen2.5-Omni-7B.
 See examples/online_serving/qwen2_5_omni/README.md
@@ -42,7 +45,7 @@ test_params = [
 common_args = ["python", os.path.join(example_dir, "openai_chat_completion_client_for_multimodal_generation.py")]
 
 
-@hardware_test(res={"cuda": "L4", "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
+@hardware_test(res={"cuda": ["L4", "B200"], "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_send_multimodal_request_001(omni_server) -> None:
     command = common_args + [
@@ -77,7 +80,7 @@ def test_send_multimodal_request_001(omni_server) -> None:
     # TODO: Verify the E2E latency after confirmation baseline.
 
 
-@hardware_test(res={"cuda": "L4", "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
+@hardware_test(res={"cuda": ["L4", "B200"], "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_send_multimodal_request_002(omni_server) -> None:
     command = common_args + [
@@ -112,7 +115,7 @@ def test_send_multimodal_request_002(omni_server) -> None:
     # TODO: Verify the E2E latency after confirmation baseline.
 
 
-@hardware_test(res={"cuda": "L4", "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
+@hardware_test(res={"cuda": ["L4", "B200"], "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_send_multimodal_request_003(omni_server) -> None:
     command = [
@@ -137,7 +140,7 @@ def test_send_multimodal_request_003(omni_server) -> None:
     # TODO: Verify the E2E latency after confirmation baseline.
 
 
-@hardware_test(res={"cuda": "L4", "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
+@hardware_test(res={"cuda": ["L4", "B200"], "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_modality_control_001(omni_server) -> None:
     command = common_args + [
@@ -165,7 +168,7 @@ def test_modality_control_001(omni_server) -> None:
     # TODO: Verify the E2E latency after confirmation baseline.
 
 
-@hardware_test(res={"cuda": "L4", "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
+@hardware_test(res={"cuda": ["L4", "B200"], "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_modality_control_002(omni_server) -> None:
     command = common_args + [
@@ -192,7 +195,7 @@ def test_modality_control_002(omni_server) -> None:
     # TODO: Verify the E2E latency after confirmation baseline.
 
 
-@hardware_test(res={"cuda": "L4", "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
+@hardware_test(res={"cuda": ["L4", "B200"], "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_modality_control_003(omni_server) -> None:
     command = common_args + [
@@ -228,7 +231,7 @@ def test_modality_control_003(omni_server) -> None:
     # TODO: Verify the E2E latency after confirmation baseline.
 
 
-@hardware_test(res={"cuda": "L4", "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
+@hardware_test(res={"cuda": ["L4", "B200"], "rocm": "MI325"}, num_cards={"cuda": 4, "rocm": 2})
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_stream_001(omni_server) -> None:
     command = common_args + [

tests/examples/online_serving/test_qwen3_omni.py

@@ -1,3 +1,6 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
 """
 Online serving tests: Qwen3-Omni.
 See examples/online_serving/qwen3_omni/README.md
@@ -41,7 +44,7 @@ test_params = [
 common_args = ["python", os.path.join(example_dir, "openai_chat_completion_client_for_multimodal_generation.py")]
 
 
-@hardware_test(res={"cuda": "H100", "rocm": "MI325"}, num_cards=2)
+@hardware_test(res={"cuda": ["H100", "B200"], "rocm": "MI325"}, num_cards=2)
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_send_multimodal_request_001(omni_server) -> None:
     command = common_args + [
@@ -69,7 +72,7 @@ def test_send_multimodal_request_001(omni_server) -> None:
     # TODO: Verify the E2E latency after confirmation baseline.
 
 
-@hardware_test(res={"cuda": "H100", "rocm": "MI325"}, num_cards=2)
+@hardware_test(res={"cuda": ["H100", "B200"], "rocm": "MI325"}, num_cards=2)
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_send_multimodal_request_002(omni_server) -> None:
     command = common_args + [
@@ -100,7 +103,7 @@ def test_send_multimodal_request_002(omni_server) -> None:
     # TODO: Verify the E2E latency after confirmation baseline.
 
 
-@hardware_test(res={"cuda": "H100", "rocm": "MI325"}, num_cards=2)
+@hardware_test(res={"cuda": ["H100", "B200"], "rocm": "MI325"}, num_cards=2)
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_send_multimodal_request_003(omni_server) -> None:
     command = ["bash", os.path.join(example_dir, "qwen3_omni/run_curl_multimodal_generation.sh"), "use_image"]
@@ -115,7 +118,7 @@ def test_send_multimodal_request_003(omni_server) -> None:
     # TODO: Verify the E2E latency after confirmation baseline.
 
 
-@hardware_test(res={"cuda": "H100", "rocm": "MI325"}, num_cards=2)
+@hardware_test(res={"cuda": ["H100", "B200"], "rocm": "MI325"}, num_cards=2)
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_modality_control_001(omni_server) -> None:
     command = common_args + [
@@ -137,7 +140,7 @@ def test_modality_control_001(omni_server) -> None:
     # TODO: Verify the E2E latency after confirmation baseline.
 
 
-@hardware_test(res={"cuda": "H100", "rocm": "MI325"}, num_cards=2)
+@hardware_test(res={"cuda": ["H100", "B200"], "rocm": "MI325"}, num_cards=2)
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_modality_control_002(omni_server) -> None:
     command = common_args + [
@@ -159,7 +162,7 @@ def test_modality_control_002(omni_server) -> None:
     # TODO: Verify the E2E latency after confirmation baseline.
 
 
-@hardware_test(res={"cuda": "H100", "rocm": "MI325"}, num_cards=2)
+@hardware_test(res={"cuda": ["H100", "B200"], "rocm": "MI325"}, num_cards=2)
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_modality_control_003(omni_server) -> None:
     command = common_args + [
@@ -189,7 +192,7 @@ def test_modality_control_003(omni_server) -> None:
     # TODO: Verify the E2E latency after confirmation baseline.
 
 
-@hardware_test(res={"cuda": "H100", "rocm": "MI325"}, num_cards=2)
+@hardware_test(res={"cuda": ["H100", "B200"], "rocm": "MI325"}, num_cards=2)
 @pytest.mark.parametrize("omni_server", test_params, indirect=True)
 def test_stream_001(omni_server) -> None:
     command = common_args + [

tests/examples/online_serving/test_text_to_image.py

@@ -1,3 +1,6 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
 """
 Online serving tests: text-to-image.
 See examples/online_serving/text_to_image/README.md
@@ -18,7 +21,7 @@ from tests.helpers.assertions import assert_image_valid
 from tests.helpers.mark import hardware_marks
 from tests.helpers.runtime import OmniServer, OmniServerParams
 
-pytestmark = [pytest.mark.full_model, pytest.mark.example, *hardware_marks(res={"cuda": "H100"})]
+pytestmark = [pytest.mark.full_model, pytest.mark.example, *hardware_marks(res={"cuda": ["H100", "B200"]})]
 
 T2I_ONLINE_CLIENT = EXAMPLES / "online_serving" / "text_to_image" / "openai_chat_client.py"
 EXAMPLE_OUTPUT_SUBFOLDER = "example_online_t2i"