Skip to content

[None][test] Consolidate ssm dis-agg E2E Tests #17680

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
bo-nv merged 4 commits into from
Aug 27, 2026
Merged

[None][test] Consolidate ssm dis-agg E2E Tests #17680

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
3f27835
[None][test] Consolidate ssm dis-agg E2E Tests
bo-nv Aug 14, 2026
88b8ac4
fix
bo-nv Aug 17, 2026
609fce8
fix test def
bo-nv Aug 24, 2026
7915a21
fix test def
bo-nv Aug 25, 2026
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
19 changes: 10 additions & 9 deletions tests/integration/defs/accuracy/test_disaggregated_serving.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2204,20 +2204,15 @@ def _make_configs(self, use_py_transceiver: bool = False):
@pytest.mark.parametrize(
"mtp_nextn,block_reuse,use_py_transceiver",
[
(0, False, False),
(0, False, True),
(3, True, False),
(3, True, True),
Comment thread
bo-nv marked this conversation as resolved.
],
ids=[
"mtp_nextn=0-block_reuse=False-use_py_transceiver=False",
"mtp_nextn=0-block_reuse=False-use_py_transceiver=True",
"mtp_nextn=3-block_reuse=True-use_py_transceiver=False",
"mtp_nextn=3-block_reuse=True-use_py_transceiver=True",
Comment thread
coderabbitai[bot] marked this conversation as resolved.
],
)
def test_auto_dtype(self, mtp_nextn, block_reuse, use_py_transceiver):
if use_py_transceiver and block_reuse:
pytest.skip("Python transceiver does not support block reuse")

ctx_cfg, gen_cfg, disagg_cfg = self._make_configs(use_py_transceiver)
if mtp_nextn > 0:
spec = {"decoding_type": "MTP", "max_draft_len": mtp_nextn}
Expand All @@ -2226,14 +2221,20 @@ def test_auto_dtype(self, mtp_nextn, block_reuse, use_py_transceiver):
if block_reuse:
ctx_cfg["kv_cache_config"]["enable_block_reuse"] = True
gen_cfg["kv_cache_config"]["enable_block_reuse"] = True
ctx_cfg["kv_cache_config"]["mamba_state_config"] = {
"periodic_snapshot_interval": 256
}
gen_cfg["kv_cache_config"]["mamba_state_config"] = {
"periodic_snapshot_interval": 256
}
with launch_disaggregated_llm(disagg_cfg, ctx_cfg, gen_cfg,
self.MODEL_PATH) as llm:
run_accuracy_test(llm, self.MODEL_NAME, ["GSM8K"])

@pytest.mark.skip_less_device(8)
def test_ctx_dp2_gen_tp4(self):
ctx_cfg, gen_cfg, disagg_cfg = self._make_configs(
use_py_transceiver=False)
use_py_transceiver=True)
# corner case: max_batch_size = 1 + dp for ctx to check if dp dummy requests are handled correctly
ctx_cfg["max_batch_size"] = 1
ctx_cfg["enable_attention_dp"] = True
Expand Down Expand Up @@ -2315,7 +2316,7 @@ def _make_configs(self, use_py_transceiver: bool):
return ctx_server_config, gen_server_config, disaggregated_server_config

@pytest.mark.skip_less_device(8)
@parametrize_with_ids("use_py_transceiver", [True, False])
@parametrize_with_ids("use_py_transceiver", [True])
def test_auto_dtype(self, use_py_transceiver, mocker):
mocker.patch.object(GSM8K, "MAX_OUTPUT_LEN", 512)
ctx_cfg, gen_cfg, disagg_cfg = self._make_configs(use_py_transceiver)
Expand Down
4 changes: 1 addition & 3 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 @@ -51,11 +51,9 @@ accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_tp_pp_symm
accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_tp_pp_symmetric[MMLU-tp1pp2]
accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_tp_pp_symmetric[MMLU-tp2pp1]
accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_tp_pp_symmetric[MMLU-tp2pp2]
accuracy/test_disaggregated_serving.py::TestNemotron3Super120B::test_auto_dtype[mtp_nextn=0-block_reuse=False-use_py_transceiver=False]
accuracy/test_disaggregated_serving.py::TestNemotron3Super120B::test_auto_dtype[mtp_nextn=0-block_reuse=False-use_py_transceiver=True]
accuracy/test_disaggregated_serving.py::TestNemotron3Super120B::test_auto_dtype[mtp_nextn=3-block_reuse=True-use_py_transceiver=False]
accuracy/test_disaggregated_serving.py::TestNemotron3Super120B::test_auto_dtype[mtp_nextn=3-block_reuse=True-use_py_transceiver=True]
Comment thread
bo-nv marked this conversation as resolved.
accuracy/test_disaggregated_serving.py::TestNemotron3Super120B::test_ctx_dp2_gen_tp4
accuracy/test_disaggregated_serving.py::TestQwen3NextInstruct::test_auto_dtype[use_py_transceiver=False]
accuracy/test_disaggregated_serving.py::TestQwen3NextInstruct::test_auto_dtype[use_py_transceiver=True]
accuracy/test_disaggregated_serving.py::TestQwen3_30B_A3B::test_mixed_ctx_gen_model[ctxpp2gentp2]
accuracy/test_disaggregated_serving.py::TestQwen3_8B::test_auto_dtype[False-False]
Expand Down
4 changes: 1 addition & 3 deletions tests/integration/test_lists/test-db/l0_dgx_b200.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -194,16 +194,14 @@ l0_dgx_b200:
- accuracy/test_llm_api_pytorch.py::TestNemotronV3Ultra::test_nvfp4_8gpus[attention_dp_on-cutedsl] TIMEOUT (60)
- accuracy/test_llm_api_pytorch.py::TestNemotronV3Ultra::test_nvfp4_8gpus[attention_dp_off-trtllm] TIMEOUT (60)
- accuracy/test_disaggregated_serving.py::TestNemotron3Super120B::test_auto_dtype[mtp_nextn=0-block_reuse=False-use_py_transceiver=True] TIMEOUT (60)
- accuracy/test_disaggregated_serving.py::TestNemotron3Super120B::test_auto_dtype[mtp_nextn=3-block_reuse=True-use_py_transceiver=False] TIMEOUT (60)
- accuracy/test_disaggregated_serving.py::TestNemotron3Super120B::test_auto_dtype[mtp_nextn=0-block_reuse=False-use_py_transceiver=False] TIMEOUT (60)
- accuracy/test_disaggregated_serving.py::TestNemotron3Super120B::test_auto_dtype[mtp_nextn=3-block_reuse=True-use_py_transceiver=True] TIMEOUT (60)
- accuracy/test_disaggregated_serving.py::TestGLM52NVFP4::test_nvfp4_nixl[cache_mgr_v1] TIMEOUT (60)
# DeepSeek-V4 EPLB pre-merge sanity (uncomment once DeepSeek-V4-Flash/Flash-Base
# checkpoints are staged under llm_models_root()).
# - accuracy/test_llm_api_pytorch.py::TestDeepSeekV4Flash::test_nvfp4_4gpus_static_eplb[moe_backend=TRTLLM] TIMEOUT (120)
# - accuracy/test_llm_api_pytorch.py::TestDeepSeekV4FlashBase::test_fp8_4gpus_static_eplb[moe_backend=DEEPGEMM] TIMEOUT (120)
- accuracy/test_disaggregated_serving.py::TestNemotron3Super120B::test_ctx_dp2_gen_tp4 TIMEOUT (60)
- accuracy/test_disaggregated_serving.py::TestQwen3NextInstruct::test_auto_dtype[use_py_transceiver=True] TIMEOUT (60)
- accuracy/test_disaggregated_serving.py::TestQwen3NextInstruct::test_auto_dtype[use_py_transceiver=False] TIMEOUT (60)
# ------------- VisualGen multi-GPU tests ---------------
- unittest/_torch/visual_gen/multi_gpu/test_attn2d_attention.py
- unittest/_torch/visual_gen/multi_gpu/test_cosmos3_transformer_parallel.py
Expand Down
4 changes: 0 additions & 4 deletions tests/integration/test_lists/waives.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,10 +8,6 @@ accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_tp_pp_symm
accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_tp_pp_symmetric[GSM8K-tp2pp2] SKIP (https://nvbugs/6428069)
accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_tp_pp_symmetric[MMLU-tp2pp1] SKIP (https://nvbugs/6611817)
accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_tp_pp_symmetric[MMLU-tp2pp2] SKIP (https://nvbugs/6428069)
accuracy/test_disaggregated_serving.py::TestNemotron3Super120B::test_auto_dtype[mtp_nextn=0-block_reuse=False-use_py_transceiver=True] SKIP (https://nvbugs/6644482)
accuracy/test_disaggregated_serving.py::TestNemotron3Super120B::test_auto_dtype[mtp_nextn=3-block_reuse=True-use_py_transceiver=False] SKIP (https://nvbugs/6561563)
accuracy/test_disaggregated_serving.py::TestNemotron3Super120B::test_ctx_dp2_gen_tp4 SKIP (https://nvbugs/6577550)
accuracy/test_disaggregated_serving.py::TestQwen3NextInstruct::test_auto_dtype[use_py_transceiver=False] SKIP (https://nvbugs/6535790)
accuracy/test_disaggregated_serving.py::TestQwen3_30B_A3B::test_mixed_ctx_gen_model[ctxpp2gentp2] SKIP (https://nvbugs/5748664)
accuracy/test_dwdp_aggregated.py::TestDwdpAggDeepSeekV3Lite::test_dwdp_agg_accuracy[mode_a_uniform_contention_opt] SKIP (https://nvbugs/6644489)
accuracy/test_llm_api_autodeploy.py::TestNemotronNanoV3::test_accuracy[nvfp4-4-attn_dp_off-trtllm] SKIP (https://nvbugs/6367792)
Expand Down
Loading