Skip to content

[TRTLLM-13629][test] Optimize MoE CI test-db #15624

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
xxi-nv merged 1 commit into from
Jun 28, 2026
Merged

[TRTLLM-13629][test] Optimize MoE CI test-db #15624

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: 2 additions & 0 deletions jenkins/L0_Test.groovy
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4467,6 +4467,7 @@ def launchTestJobs(pipeline, testFilter)
"DGX_H100-4_GPUs-PyTorch-Ray-1": ["auto:dgx-h100-x4", "l0_dgx_h100", 1, 1, 4],
"DGX_H100-4_GPUs-AutoDeploy-1": ["auto:dgx-h100-x4", "l0_dgx_h100", 1, 1, 4],
"DGX_H100-4_GPUs-AutoDeploy-Post-Merge-1": ["auto:dgx-h100-x4", "l0_dgx_h100", 1, 1, 4],
"DGX_H100-4_GPUs-PyTorch-Post-Merge-1": ["auto:dgx-h100-x4", "l0_dgx_h100", 1, 1, 4],
"DGX_B200-PyTorch-1": ["auto:dgx-b200-flex", "l0_b200", 1, 9, 1, 1, true],
"DGX_B200-PyTorch-2": ["auto:dgx-b200-flex", "l0_b200", 2, 9, 1, 1, true],
"DGX_B200-PyTorch-3": ["auto:dgx-b200-flex", "l0_b200", 3, 9, 1, 1, true],
Expand Down Expand Up @@ -4497,6 +4498,7 @@ def launchTestJobs(pipeline, testFilter)
"DGX_B200-4_GPUs-Verl-Post-Merge-1": ["auto:dgx-b200-flex", "l0_verl", 1, 1, 4, 1, true],
"B300-PyTorch-1": ["auto:dgx-b300-flex", "l0_b300", 1, 2, 1, 1, true],
"B300-PyTorch-2": ["auto:dgx-b300-flex", "l0_b300", 2, 2, 1, 1, true],
"B300-PyTorch-Post-Merge-1": ["auto:dgx-b300-flex", "l0_b300", 1, 1, 1, 1, true],
"DGX_B300-4_GPUs-PyTorch-1": ["auto:dgx-b300-flex", "l0_dgx_b300", 1, 1, 4, 1, true],
"DGX_B300-4_GPUs-PyTorch-Post-Merge-1": ["auto:dgx-b300-flex", "l0_dgx_b300", 1, 2, 4, 1, true],
"DGX_B300-4_GPUs-PyTorch-Post-Merge-2": ["auto:dgx-b300-flex", "l0_dgx_b300", 2, 2, 4, 1, true],
Expand Down
27 changes: 14 additions & 13 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 @@ -33,9 +33,6 @@ l0_b200:
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_nvfp4[moe_backend=CUTLASS-mtp_nextn=0-fp8kv=True-attention_dp=False-cuda_graph=True-overlap_scheduler=True-torch_compile=True]
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_nvfp4[moe_backend=CUTLASS-mtp_nextn=0-fp8kv=True-attention_dp=True-cuda_graph=True-overlap_scheduler=True-torch_compile=False]
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_nvfp4[moe_backend=TRTLLM-mtp_nextn=0-fp8kv=True-attention_dp=False-cuda_graph=True-overlap_scheduler=True-torch_compile=False]
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_nvfp4[moe_backend=CUTLASS-mtp_nextn=2-fp8kv=True-attention_dp=False-cuda_graph=True-overlap_scheduler=True-torch_compile=False]
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_nvfp4[moe_backend=TRTLLM-mtp_nextn=2-fp8kv=True-attention_dp=False-cuda_graph=True-overlap_scheduler=True-torch_compile=False]
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_nvfp4[moe_backend=CUTEDSL-mtp_nextn=2-fp8kv=True-attention_dp=True-cuda_graph=True-overlap_scheduler=True-torch_compile=False]
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_no_kv_cache_reuse[quant_dtype=none-mtp_nextn=2-fp8kv=False-attention_dp=True-cuda_graph=True-overlap_scheduler=True]
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_no_kv_cache_reuse[quant_dtype=nvfp4-mtp_nextn=0-fp8kv=True-attention_dp=True-cuda_graph=True-overlap_scheduler=True]
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_chunked_prefill[quant_dtype=none-kv_cache_reuse=True-fp8kv=False-overlap_scheduler=True]
Expand Down Expand Up @@ -122,19 +119,10 @@ l0_b200:
- unittest/_torch/modules/test_moe_host_sharer.py
- unittest/_torch/modules/fused_moe/test_deepgemm_fused_gather_finalize.py
- unittest/_torch/modules/fused_moe/test_deepgemm_fused_expand_quant.py
# ------------- legacy MoE tests ---------------
- unittest/_torch/modules/test_fused_moe.py
# ------------- MoE: test_moe_backend (by backend) ---------------
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "CUTLASS"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "TRTLLM"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "CUTEDSL and not MEGAMOE_CUTEDSL"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "DEEPGEMM and not MEGAMOE_DEEPGEMM"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "DENSEGEMM"
- unittest/_torch/modules/moe/test_moe_backend.py::test_trtllm_bf16_unquantized_moe
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "MEGAMOE_CUTEDSL"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "MEGAMOE_DEEPGEMM"
# ------------- MoE: test_single_gpu (by backend) ---------------
- unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_single_gpu -k "CUTLASS"
- unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_single_gpu -k "CUTLASS and not None"
- unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_single_gpu -k "TRTLLM"
- unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_single_gpu -k "CUTEDSL and not MEGAMOE_CUTEDSL"
- unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_single_gpu -k "DEEPGEMM and not MEGAMOE_DEEPGEMM"
Expand Down Expand Up @@ -363,6 +351,19 @@ l0_b200:
- examples/visual_gen/test_visual_gen.py::test_wan22_t2v_lpips_against_golden
- visual_gen/test_visual_gen_benchmark.py::test_offline_benchmark
- visual_gen/test_visual_gen_benchmark.py::test_online_benchmark[openai-videos]
# ---- moved to post-merge (MoE CI optimization) ----
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_nvfp4[moe_backend=CUTLASS-mtp_nextn=2-fp8kv=True-attention_dp=False-cuda_graph=True-overlap_scheduler=True-torch_compile=False]
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_nvfp4[moe_backend=TRTLLM-mtp_nextn=2-fp8kv=True-attention_dp=False-cuda_graph=True-overlap_scheduler=True-torch_compile=False]
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_nvfp4[moe_backend=CUTEDSL-mtp_nextn=2-fp8kv=True-attention_dp=True-cuda_graph=True-overlap_scheduler=True-torch_compile=False]
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "CUTLASS"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "TRTLLM"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "CUTEDSL and not MEGAMOE_CUTEDSL"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "DEEPGEMM and not MEGAMOE_DEEPGEMM"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "DENSEGEMM"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "MEGAMOE_CUTEDSL"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "MEGAMOE_DEEPGEMM"
# ---- non-quantized (quant=None) moved to post-merge ----
- unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_single_gpu -k "CUTLASS and None"
# ------------- AutoDeploy Backend Stages ---------------
- condition:
ranges:
Expand Down
26 changes: 20 additions & 6 deletions tests/integration/test_lists/test-db/l0_b300.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,13 +34,7 @@ l0_b300:
- unittest/_torch/modules/test_moe_routing.py
- unittest/_torch/modules/fused_moe/test_deepgemm_fused_gather_finalize.py
- unittest/_torch/modules/fused_moe/test_deepgemm_fused_expand_quant.py
# ------------- legacy MoE tests ---------------
- unittest/_torch/modules/test_fused_moe.py
# ------------- MoE: test_moe_backend (by backend) ---------------
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "CUTLASS"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "TRTLLM"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "CUTEDSL"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "DEEPGEMM"
- unittest/_torch/modules/moe/test_moe_backend.py::test_trtllm_bf16_unquantized_moe
# ------------- MoE: test_single_gpu (specific quant per backend) ---------------
# CUTLASS backend: FP8, NVFP4, W4A8_MXFP4_MXFP8, W8A16
Expand All @@ -64,3 +58,23 @@ l0_b300:
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_chunked_prefill[quant_dtype=fp8-kv_cache_reuse=True-fp8kv=True-overlap_scheduler=True] # Cover nvbugs 6084445
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_bfloat16_flashinfer[enable_chunked_prefill=False]
- accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_bfloat16_flashinfer[enable_chunked_prefill=True]
- condition:
ranges:
system_gpu_count:
gte: 1
lte: 1
wildcards:
gpu:
- '*gb110*'
- '*b300*'
linux_distribution_name: ubuntu*
cpu: x86_64
terms:
stage: post_merge
backend: pytorch
tests:
# ---- moved to post-merge (MoE CI optimization) ----
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "CUTLASS"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "TRTLLM"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "CUTEDSL"
- unittest/_torch/modules/moe/test_moe_backend.py::test_moe_backend -k "DEEPGEMM"
Comment thread
xxi-nv marked this conversation as resolved.
Loading
Loading