Skip to content

[https://nvbugs/5777041][fix] fix AutoDeploy ep sharding test #10460

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

Open
lucaslie wants to merge 1 commit into
base: main
Choose a base branch
from
+3 −7
Open

[https://nvbugs/5777041][fix] fix AutoDeploy ep sharding test #10460

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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: 0 additions & 2 deletions tests/integration/test_lists/waives.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -491,8 +491,6 @@ triton_server/test_triton.py::test_llava_onevision[llava_onevision] SKIP (https:
triton_server/test_triton.py::test_gpt_ib_lad[gpt-ib-lad] SKIP (https://nvbugs/5775223)
accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B::test_nvfp4[dep4_latency_moe_trtllm-torch_compile=True] SKIP (https://nvbugs/5740377)
unittest/_torch/modules/test_fused_moe.py::test_fused_moe_fp8_blockwise_cute_dsl_multi_gpu[MoEWeightLoadingMode.FUSED_GATE_UP_PROJ-DefaultMoeRoutingMethod-1] SKIP (https://nvbugs/5775256)
unittest/_torch/auto_deploy/unit/multigpu/transformations/library/test_ep_sharding.py::test_ep_shard[3-2] SKIP (https://nvbugs/5777041)
unittest/_torch/auto_deploy/unit/multigpu/transformations/library/test_ep_sharding.py::test_ep_shard[8-2] SKIP (https://nvbugs/5777041)
disaggregated/test_disaggregated.py::test_disaggregated_benchmark_on_diff_backends[llama-3.1-8b-instruct-hf-fp8] SKIP (https://nvbugs/5769890)
disaggregated/test_disaggregated.py::test_disaggregated_benchmark_on_diff_backends[DeepSeek-V3-Lite-bf16] SKIP (https://nvbugs/5769890)
disaggregated/test_disaggregated.py::test_disaggregated_benchmark_on_diff_backends[llama-v3-8b-hf] SKIP (https://nvbugs/5769890,https://nvbugs/5748683)
Expand Down
8 changes: 3 additions & 5 deletions tests/unittest/_torch/auto_deploy/unit/multigpu/transformations/library/test_ep_sharding.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,10 +31,6 @@ def _run_ep_shard_job(num_experts: int, rank: int, world_size: int) -> None:
).to(device=device, dtype=torch.bfloat16)
x = model.get_input(device=device, dtype=torch.bfloat16)

if world_size > num_experts:
print(f"world_size {world_size} > num_experts {num_experts}, skipping test")
return

def _get_expected_num_params(rank: int, world_size: int, num_p_og: int) -> int:
if world_size <= 1:
return num_p_og
Expand Down Expand Up @@ -141,9 +137,11 @@ def _run_pattern_detection_job(num_experts: int, rank: int, world_size: int) ->
run_sharding_pattern_detection_test(detected_transformations, expected_transformations)


@pytest.mark.parametrize("device_count", get_device_counts())
@pytest.mark.parametrize("device_count", get_device_counts([2, 8]))
@pytest.mark.parametrize("num_experts", [3, 8])
def test_ep_shard(device_count: int, num_experts: int):
if device_count > num_experts:
pytest.skip(f"world_size {device_count} > num_experts {num_experts}")
dist_common.spawn_multiprocess_job(
job=partial(_run_ep_shard_job, num_experts),
size=device_count,
Expand Down
Loading