Skip to content

[https://nvbugs/5325296][fix] Enable relaxed acceptance test on Blackwell #8709

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
merged 8 commits into from
Oct 31, 2025
Merged

[https://nvbugs/5325296][fix] Enable relaxed acceptance test on Blackwell #8709

Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
34cd672
Enable relaxed acceptance test on Blackwell
Barry-Delaney Oct 28, 2025
57f5ba5
Fix format
Barry-Delaney Oct 28, 2025
1778fef
Lower KV fraction for Blackwell
Barry-Delaney Oct 29, 2025
43c4ea1
Merge branch 'release/1.1' into user/barry/fix_sm100_r1_ci
Barry-Delaney Oct 29, 2025
f0363ca
Merge branch 'release/1.1' into user/barry/fix_sm100_r1_ci
Barry-Delaney Oct 29, 2025
2673902
Merge branch 'release/1.1' into user/barry/fix_sm100_r1_ci
Barry-Delaney Oct 30, 2025
04de7da
Merge branch 'release/1.1' into user/barry/fix_sm100_r1_ci
Barry-Delaney Oct 31, 2025
f077b6d
Merge branch 'release/1.1' into user/barry/fix_sm100_r1_ci
Barry-Delaney Oct 31, 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
14 changes: 8 additions & 6 deletions tests/integration/defs/test_e2e.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,10 +31,10 @@
from .common import (PluginOptions, convert_weights, get_mmlu_accuracy,
prune_checkpoint, quantize_data, refit_model,
venv_check_call)
from .conftest import (get_device_count, llm_models_root, skip_no_sm120,
skip_nvlink_inactive, skip_post_blackwell, skip_pre_ada,
skip_pre_blackwell, skip_pre_hopper, tests_path,
unittest_path)
from .conftest import (get_device_count, get_sm_version, llm_models_root,
skip_no_sm120, skip_nvlink_inactive, skip_post_blackwell,
skip_pre_ada, skip_pre_blackwell, skip_pre_hopper,
tests_path, unittest_path)

sys.path.append(os.path.join(str(tests_path()), '/../examples/apps'))

Expand Down Expand Up @@ -2184,7 +2184,6 @@ def test_ptp_quickstart_advanced_deepseek_r1_8gpus(llm_root, llm_venv,
_check_mem_usage(running_log, [106.3, 0, 0, 0], 8)


@skip_post_blackwell
@pytest.mark.skip_less_device_memory(110000)
@pytest.mark.skip_less_device(8)
@pytest.mark.parametrize("model_name,model_path", [
Expand All @@ -2195,6 +2194,7 @@ def test_relaxed_acceptance_quickstart_advanced_deepseek_r1_8gpus(
llm_root, llm_venv, model_name, model_path):
print(f"Testing {model_name}.")
example_root = Path(os.path.join(llm_root, "examples", "llm-api"))
is_blackwell = get_sm_version() > 90
with tempfile.NamedTemporaryFile(mode='w+t',
suffix=f".{model_name}.log",
dir="./",
Expand All @@ -2208,7 +2208,7 @@ def test_relaxed_acceptance_quickstart_advanced_deepseek_r1_8gpus(
"--moe_ep_size=8",
"--tp_size=8",
"--use_cuda_graph",
f"--kv_cache_fraction={_MEM_FRACTION_95}",
f"--kv_cache_fraction={_MEM_FRACTION_50 if is_blackwell else _MEM_FRACTION_95}",
"--max_batch_size=1",
"--max_seq_len=3000",
"--disable_kv_cache_reuse",
Expand All @@ -2221,6 +2221,8 @@ def test_relaxed_acceptance_quickstart_advanced_deepseek_r1_8gpus(
"--relaxed_delta=0.5",
"--enable_attention_dp",
"--use_one_model",
"--moe_backend",
"DEEPGEMM" if is_blackwell else "CUTLASS",
],
stdout=running_log)
_check_mem_usage(running_log, [85.6, 0, 0, 0], 8)
Expand Down