Skip to content

[TRTLLM-12365][ci] Dedup AutoDeploy unit tests on B200 #13593

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
QiJune merged 4 commits into from
May 1, 2026
Merged

[TRTLLM-12365][ci] Dedup AutoDeploy unit tests on B200 #13593

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
6c67e30
Dedup AutoDeploy unit tests on B200
QiJune Apr 29, 2026
7c822c3
update
QiJune Apr 29, 2026
3121617
update
QiJune Apr 29, 2026
5fd43cb
[None][ci] Restore B-tier coverage flagged in review
QiJune Apr 30, 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
24 changes: 17 additions & 7 deletions tests/integration/test_lists/test-db/l0_b200.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -313,13 +313,23 @@ l0_b200:
- accuracy/test_llm_api_autodeploy.py::TestNemotronSuperV3::test_accuracy[nvfp4-1-attn_dp_off-trtllm]
- accuracy/test_llm_api_autodeploy.py::TestNemotronSuperV3::test_functional_small[bf16]
- accuracy/test_llm_api_autodeploy.py::TestNemotronSuperV3::test_functional_small[fp8]
- unittest/auto_deploy/singlegpu/compile
- unittest/auto_deploy/singlegpu/custom_ops
- unittest/auto_deploy/singlegpu/models
- unittest/auto_deploy/singlegpu/shim
- unittest/auto_deploy/singlegpu/smoke
- unittest/auto_deploy/singlegpu/transformations
- unittest/auto_deploy/singlegpu/utils
# ------------- AutoDeploy unit tests ---------------
# B-tier keeps only Blackwell-meaningful tests (NVFP4 / FP4 / TRTLLM-Gen /
# SM100+-gated). HW-agnostic tests (compile, models, shim, utils, plus most
# custom_ops, smoke, transformations files) and pure-FP8 tests are covered
# on Hopper (l0_h100.yml) and not duplicated here.
- unittest/auto_deploy/singlegpu/custom_ops/attention/test_triton_paged_attention.py::TestSDPADispatch
- unittest/auto_deploy/singlegpu/custom_ops/moe/test_ad_moe_op.py
- unittest/auto_deploy/singlegpu/custom_ops/moe/test_trtllm_moe.py
- unittest/auto_deploy/singlegpu/custom_ops/quantization/test_quant.py
Comment thread
QiJune marked this conversation as resolved.
Resolved
- unittest/auto_deploy/singlegpu/smoke/test_ad_build_small_single.py -k "Nemotron-3-Nano-30B-A3B-FP8 or Nemotron-Nano-3-30B-A3.5B-dev or Llama-4-Scout"
- unittest/auto_deploy/singlegpu/transformations/library/test_fuse_relu2_quant_nvfp4.py
- unittest/auto_deploy/singlegpu/transformations/library/test_moe_fusion.py
- unittest/auto_deploy/singlegpu/transformations/library/test_nvfp4_swiglu.py
- unittest/auto_deploy/singlegpu/transformations/library/test_quant_fusion.py
- unittest/auto_deploy/singlegpu/transformations/library/test_quant_moe.py
- unittest/auto_deploy/singlegpu/transformations/library/test_quantization.py
- unittest/auto_deploy/singlegpu/transformations/library/test_trtllm_gen_diag.py
# ------------- AutoDeploy Perf Sanity ---------------
- perf/test_perf_sanity.py::test_e2e[aggr_upload-super_ad_blackwell-super_ad_ws1_1k1k] TIMEOUT (120)

Expand Down
8 changes: 7 additions & 1 deletion 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 @@ -322,7 +322,13 @@ l0_dgx_b200:
tests:
- unittest/auto_deploy/multigpu/custom_ops
- unittest/auto_deploy/multigpu/smoke
- unittest/auto_deploy/multigpu/transformations
# B-tier keeps the NVFP4 MoE TP sharding test (SM100+ gated) and the
# strategy_auto parametrization of the allreduce-residual-rmsnorm fusion test
# (covers AUTO -> NVLS dispatch on B-tier, structurally different from H100).
# The other multigpu/transformations tests are HW-agnostic and covered on
# Hopper (l0_dgx_h100.yml).
- unittest/auto_deploy/multigpu/transformations/library/test_tp_sharding.py::test_moe_tp_shard_nvfp4
Comment thread
QiJune marked this conversation as resolved.
Resolved
- unittest/auto_deploy/multigpu/transformations/library/test_allreduce_residual_rmsnorm_fusion.py -k "strategy_auto"
- accuracy/test_llm_api_autodeploy.py::TestLlama3_1_8B::test_auto_dtype[trtllm-False-4]
- accuracy/test_llm_api_autodeploy.py::TestNemotronNanoV3::test_accuracy[fp8-4-trtllm]
- accuracy/test_llm_api_autodeploy.py::TestNemotronNanoV3::test_accuracy[nvfp4-4-trtllm]
Expand Down
Loading