Skip to content

[https://nvbugs/5494698][fix] skip gemma3 27b on blackwell #7505

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
xinhe-nv merged 2 commits into from
Sep 10, 2025
Merged

[https://nvbugs/5494698][fix] skip gemma3 27b on blackwell #7505

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
32f6b3a
skip gemma3 27b on blackwell
xinhe-nv Sep 4, 2025
4d5cff3
fix test_trtllm_serve_multimodal_benchmark
xinhe-nv Sep 4, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions tests/integration/defs/accuracy/test_llm_api_pytorch.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -970,6 +970,8 @@ def test_fp8(self):
pytest.skip("FP8 pre-quantized Ministral-8B model not available")


@skip_post_blackwell
@skip_pre_hopper
class TestGemma3_27BInstruct(LlmapiAccuracyTestHarness):
MODEL_NAME = "google/gemma-3-27b-it"
MODEL_PATH = f"{llm_models_root()}/gemma/gemma-3-27b-it/"
Expand Down
8 changes: 4 additions & 4 deletions tests/integration/defs/test_e2e.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2333,8 +2333,8 @@ def test_ptp_quickstart_advanced_mixed_precision(llm_root, llm_venv):
marks=pytest.mark.skip_less_device_memory(80000)),
pytest.param("gemma-3-27b-it",
"gemma/gemma-3-27b-it",
marks=(skip_post_blackwell,
pytest.mark.skip_less_device_memory(80000))),
marks=(pytest.mark.skip_less_device_memory(80000),
skip_post_blackwell)),
])
def test_ptp_quickstart_multimodal(llm_root, llm_venv, model_name, model_path,
modality, use_cuda_graph):
Expand Down Expand Up @@ -2729,10 +2729,10 @@ def test_ptp_quickstart_multimodal_phi4mm(llm_root, llm_venv, modality):
@pytest.mark.skip_less_device(2)
@pytest.mark.skip_less_device_memory(80000)
@pytest.mark.parametrize("model_name,model_path", [
("mistral-small-3.1-24b-instruct", "Mistral-Small-3.1-24B-Instruct-2503"),
("Phi-4-multimodal-instruct", "multimodals/Phi-4-multimodal-instruct"),
pytest.param(
"gemma-3-27b-it", "gemma/gemma-3-27b-it", marks=skip_post_blackwell),
("mistral-small-3.1-24b-instruct", "Mistral-Small-3.1-24B-Instruct-2503"),
("Phi-4-multimodal-instruct", "multimodals/Phi-4-multimodal-instruct"),
])
def test_ptp_quickstart_multimodal_2gpu(llm_root, llm_venv, model_name,
model_path):
Expand Down
2 changes: 2 additions & 0 deletions tests/integration/test_lists/qa/llm_function_core.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -424,6 +424,8 @@ accuracy/test_llm_api_pytorch.py::TestLlama3_3_70BInstruct::test_fp8_eagle3_tp8[
accuracy/test_llm_api_pytorch.py::TestLlama3_3_70BInstruct::test_fp8_eagle3_tp8[eagle3_one_model=False]
accuracy/test_llm_api_pytorch.py::TestMistral7B::test_auto_dtype
accuracy/test_llm_api_pytorch.py::TestGemma3_1BInstruct::test_auto_dtype
accuracy/test_llm_api_pytorch.py::TestGemma3_27BInstruct::test_fp8_prequantized
accuracy/test_llm_api_pytorch.py::TestGemma3_27BInstruct::test_auto_dtype
accuracy/test_llm_api_pytorch.py::TestMistralSmall24B::test_auto_dtype
accuracy/test_llm_api_pytorch.py::TestMistralSmall24B::test_fp8
accuracy/test_llm_api_pytorch.py::TestLlama4MaverickInstruct::test_auto_dtype[tp8-cuda_graph=False]
Expand Down
2 changes: 2 additions & 0 deletions tests/integration/test_lists/qa/llm_function_core_sanity.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,8 @@ accuracy/test_disaggregated_serving.py::TestDeepSeekV3Lite::test_auto_dtype[mtp_
accuracy/test_disaggregated_serving.py::TestDeepSeekV3Lite::test_nixl_backend
accuracy/test_disaggregated_serving.py::TestGemma3_1BInstruct::test_auto_dtype[False]
accuracy/test_disaggregated_serving.py::TestGemma3_1BInstruct::test_auto_dtype[True]
accuracy/test_llm_api_pytorch.py::TestGemma3_27BInstruct::test_fp8_prequantized
accuracy/test_llm_api_pytorch.py::TestGemma3_27BInstruct::test_auto_dtype
accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_auto_dtype[False]
accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_auto_dtype[True]
accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_ctx_pp_gen_tp_asymmetric[GSM8K-gen_tp=1-ctx_pp=2]
Expand Down
21 changes: 12 additions & 9 deletions tests/unittest/llmapi/apps/_test_trtllm_serve_multimodal_benchmark.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -71,15 +71,18 @@ def vision_arena_dataset_path():


@skip_gpu_memory_less_than_80gb
@pytest.mark.parametrize("dataset_name,dataset_args", [("random_image", {
"--num-images": "1",
"--image-size": "512",
}), ("random_image", {
"--num-images": "2",
"--image-size": "512",
}), ("hf", {
"--dataset-path": vision_arena_dataset_path(),
})],
@pytest.mark.parametrize("dataset_name,dataset_args",
[("random_image", {
"--random-num-images": "1",
"--random-image-size": "512",
}),
("random_image", {
"--random-num-images": "2",
"--random-image-size": "512",
}),
("hf", {
"--dataset-path": vision_arena_dataset_path(),
})],
ids=[
"random_image-single_image",
"random_image-dual_images",
Expand Down