Skip to content

[CI] Add Decode Context Parallelism (DCP) test to CI #24487

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 7 commits into from
Sep 16, 2025
Merged

[CI] Add Decode Context Parallelism (DCP) test to CI #24487

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
e6de943
Add DCP to CI
minosfuture Sep 9, 2025
e1a4326
add comment
minosfuture Sep 9, 2025
07a0b95
address comment: combine CI tests
minosfuture Sep 9, 2025
2bf145d
Merge branch 'main' into add_dcp_ci
minosfuture Sep 10, 2025
fcddeb3
Merge branch 'main' into add_dcp_ci
minosfuture Sep 12, 2025
0d62d75
add working_dir; fix relative path
minosfuture Sep 13, 2025
8375e79
run dcp test before moe (for testing purpose)
minosfuture Sep 13, 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
17 changes: 14 additions & 3 deletions .buildkite/test-pipeline.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -898,7 +898,6 @@ steps:
commands:
- pytest -v -s distributed/test_pp_cudagraph.py
- pytest -v -s distributed/test_pipeline_parallel.py
# - pytest -v -s distributed/test_context_parallel.py # TODO: enable it on Hopper runners or add triton MLA support

- label: LoRA TP Test (Distributed) # 17 min
timeout_in_minutes: 30
Expand Down Expand Up @@ -972,9 +971,21 @@ steps:
- export VLLM_WORKER_MULTIPROC_METHOD=spawn
- pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-large.txt --tp-size=4

- label: Qwen MoE EP Test # optional
##### H200 test #####
- label: Distrubted Tests (H200) # optional
gpu: h200
optional: true
working_dir: "/vllm-workspace/"
num_gpus: 2
commands:
- pytest -v -s tests/distributed/test_context_parallel.py
- CUDA_VISIBLE_DEVICES=1,2 VLLM_ALL2ALL_BACKEND=deepep_high_throughput VLLM_USE_DEEP_GEMM=1 VLLM_LOGGING_LEVEL=DEBUG python3 examples/offline_inference/data_parallel.py --model Qwen/Qwen1.5-MoE-A2.7B --tp-size=1 --dp-size=2 --max-model-len 2048

##### B200 test #####
- label: Distributed Tests (B200) # optional
gpu: b200
optional: true
working_dir: "/vllm-workspace/"
num_gpus: 2
commands:
- CUDA_VISIBLE_DEVICES=1,2 VLLM_ALL2ALL_BACKEND=deepep_high_throughput VLLM_USE_DEEP_GEMM=1 VLLM_LOGGING_LEVEL=DEBUG python3 /vllm-workspace/examples/offline_inference/data_parallel.py --model Qwen/Qwen1.5-MoE-A2.7B --tp-size=1 --dp-size=2 --max-model-len 2048
- pytest -v -s tests/distributed/test_context_parallel.py
11 changes: 7 additions & 4 deletions tests/distributed/test_context_parallel.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -71,12 +71,13 @@ def detailed(
parallel_setups = []
for eager_mode_val in [False]:
for pp_multiplier in [1]:
for dcp_multiplier in [2, 4]:
for dcp_multiplier in [0.5, 1]:
for chunked_prefill_val in [True]:
parallel_setups.append(
ParallelSetup(tp_size=tp_base,
pp_size=pp_multiplier * pp_base,
dcp_size=dcp_multiplier * dcp_base,
dcp_size=int(dcp_multiplier *
tp_base),
eager_mode=eager_mode_val,
chunked_prefill=chunked_prefill_val))
return CPTestSettings(
Expand Down Expand Up @@ -223,7 +224,9 @@ def _compare_cp_with_tp(

CP_TEXT_GENERATION_MODELS = {
# [MLA attention only]
"deepseek-ai/DeepSeek-V2-Lite-Chat": CPTestSettings.detailed(),
"deepseek-ai/DeepSeek-V2-Lite-Chat":
[CPTestSettings.detailed(),
CPTestSettings.detailed(tp_base=2)],
}

CP_TEST_MODELS = [
Expand All @@ -238,7 +241,7 @@ def _compare_cp_with_tp(
"runner", "test_options"),
[
params for model_id, settings in CP_TEXT_GENERATION_MODELS.items()
for params in settings.iter_params(model_id)
for setting in settings for params in setting.iter_params(model_id)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

high

The list comprehension here assumes that the values in CP_TEXT_GENERATION_MODELS are always lists of CPTestSettings. This might be fragile if a new model is added with a single CPTestSettings object instead of a list. This would cause the test parameter generation to fail. To make this more robust, you could handle both cases by ensuring you're always iterating over a list.

Suggested change
for setting in settings for params in setting.iter_params(model_id)
for setting in (settings if isinstance(settings, list) else [settings]) for params in setting.iter_params(model_id)

if model_id in CP_TEST_MODELS
],
)
Expand Down