[TRTLLM-13027][ci] Relocate under-using tests to right-sized stages (#14684)

Signed-off-by: junq <22017000+QiJune@users.noreply.github.com>

Author: QiJune

jenkins/L0_Test.groovy

@@ -4053,6 +4053,7 @@ def launchTestJobs(pipeline, testFilter)
         "DGX_B200-Triton-Post-Merge-1": ["auto:dgx-b200-flex", "l0_b200", 1, 1, 1, 1, true],
         "DGX_B200-PyTorch-Post-Merge-1": ["auto:dgx-b200-flex", "l0_b200", 1, 2, 1, 1, true],
         "DGX_B200-PyTorch-Post-Merge-2": ["auto:dgx-b200-flex", "l0_b200", 2, 2, 1, 1, true],
+        "DGX_B200-2_GPUs-PyTorch-1": ["auto:dgx-b200-flex", "l0_dgx_b200", 1, 1, 2, 1, true],
         "DGX_B200-4_GPUs-PyTorch-1": ["auto:dgx-b200-flex", "l0_dgx_b200", 1, 3, 4, 1, true],
         "DGX_B200-4_GPUs-PyTorch-2": ["auto:dgx-b200-flex", "l0_dgx_b200", 2, 3, 4, 1, true],
         "DGX_B200-4_GPUs-PyTorch-3": ["auto:dgx-b200-flex", "l0_dgx_b200", 3, 3, 4, 1, true],

tests/integration/test_lists/test-db/l0_b200.yml

@@ -120,6 +120,7 @@ l0_b200:
   - unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_single_gpu -k "CUTEDSL"
   - unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_single_gpu -k "DEEPGEMM"
   - unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_single_gpu -k "DENSEGEMM"
+  - unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_single_gpu -k "MEGAMOE_DEEPGEMM"
   # ------------- MoE: FlashInfer & TRTLLM symbol collision tests ---------------
   - unittest/_torch/flashinfer/test_trtllm_flashinfer_symbol_collision.py
   # --- MoE end
@@ -307,6 +308,16 @@ l0_b200:
   - accuracy/test_llm_api_pytorch.py::TestGPTOSS::test_w4_1gpu[v1_kv_cache-True-True-trtllm-auto]
   - accuracy/test_llm_api_pytorch.py::TestGPTOSS::test_w4_1gpu[v2_kv_cache-True-True-trtllm-auto]
   - accuracy/test_llm_api_pytorch_multimodal.py::TestNanoV3Omni::test_auto_dtype[bf16]
+  # ------------- VisualGen single-GPU tests ---------------
+  - examples/test_visual_gen.py::test_visual_gen_quickstart
+  - examples/test_visual_gen.py::test_visual_gen_api_walkthrough
+  - examples/test_visual_gen.py::test_flux1_lpips_against_golden
+  - examples/test_visual_gen.py::test_flux2_lpips_against_golden
+  - examples/test_visual_gen.py::test_ltx2_lpips_against_golden
+  - examples/test_visual_gen.py::test_wan21_t2v_lpips_against_golden
+  - examples/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]
 # ------------- AutoDeploy Backend Stages ---------------
 - condition:
     ranges:

tests/integration/test_lists/test-db/l0_dgx_b200.yml

@@ -1,5 +1,30 @@
 version: 0.0.1
 l0_dgx_b200:
+- condition:
+    ranges:
+      system_gpu_count:
+        gte: 2
+        lte: 2
+    wildcards:
+      gpu:
+      - '*b200*'
+      linux_distribution_name: ubuntu*
+      cpu: x86_64
+    terms:
+      stage: pre_merge
+      backend: pytorch
+      orchestrator: mpi
+  tests:
+  - unittest/_torch/misc/test_autotuner.py::test_autotuner_distributed_strategy
+  - accuracy/test_llm_api_pytorch.py::TestQwen3_5_35B_A3B::test_bf16[tp2-CUTLASS]
+  - accuracy/test_llm_api_pytorch.py::TestQwen3_5_35B_A3B::test_bf16[tp2-TRTLLM]
+  # ------------- KV Cache V2 Scheduler IT (multi-GPU) ---------------
+  - kv_cache/test_kv_cache_v2_scheduler.py::TestKVCacheV2DSv3Lite::test_mtp_draft_tokens
+  - kv_cache/test_kv_cache_v2_scheduler.py::TestKVCacheV2DSv3Lite::test_mtp_chunked_draft_tokens
+  - kv_cache/test_kv_cache_v2_scheduler.py::TestKVCacheV2DSv3Lite::test_mtp_eviction
+  # ------------- VisualGen multi-GPU tests ---------------
+  - unittest/_torch/visual_gen/test_flux_pipeline.py::TestFluxParallelism::test_ulysses_2gpu_correctness
+  - unittest/_torch/visual_gen/test_flux_pipeline.py::TestFluxCombinedOptimizations::test_all_optimizations_combined
 - condition:
     ranges:
       system_gpu_count:
@@ -15,7 +40,6 @@ l0_dgx_b200:
       backend: pytorch
       orchestrator: mpi
   tests:
-  - unittest/_torch/misc/test_autotuner.py::test_autotuner_distributed_strategy
   - accuracy/test_llm_api_pytorch.py::TestNemotronV3Super::test_auto_dtype_4gpus[4-4-False-True-True]
   - accuracy/test_llm_api_pytorch.py::TestNemotronV3Super::test_auto_dtype_4gpus[4-4-True-True-True]
   - accuracy/test_llm_api_pytorch.py::TestNemotronV3Super::test_nvfp4_4gpu_mtp_ar TIMEOUT (60)
@@ -30,8 +54,6 @@ l0_dgx_b200:
   - accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B::test_nvfp4[dep4_latency_moe_trtllm-torch_compile=False]
   - accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B::test_nvfp4[dep4_latency_moe_cutlass-torch_compile=False]
   - accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B::test_nvfp4[dep4_latency_moe_cutlass-torch_compile=True]
-  - accuracy/test_llm_api_pytorch.py::TestQwen3_5_35B_A3B::test_bf16[tp2-CUTLASS]
-  - accuracy/test_llm_api_pytorch.py::TestQwen3_5_35B_A3B::test_bf16[tp2-TRTLLM]
   - disaggregated/test_disaggregated.py::test_disaggregated_deepseek_v3_lite_fp8_ucx[DeepSeek-V3-Lite-fp8]
   - disaggregated/test_disaggregated.py::test_disaggregated_deepseek_v3_lite_fp8_nixl[DeepSeek-V3-Lite-fp8]
   - disaggregated/test_disaggregated.py::test_disaggregated_gpt_oss_120b_harmony[gpt_oss/gpt-oss-120b]
@@ -42,10 +64,6 @@ l0_dgx_b200:
   - accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_bfloat16_4gpus_python_scheduler[ep4-mtp_nextn=2]
   - accuracy/test_llm_api_pytorch.py::TestMiniMaxM2::test_4gpus[attention_dp=False-cuda_graph=True-overlap_scheduler=True-tp_size=4-ep_size=4] TIMEOUT (60)
   - accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_bfloat16_4gpus[pp4-mtp_nextn=0-attention_dp=False-cuda_graph=False-overlap_scheduler=False-torch_compile=False] TIMEOUT (60)
-  # ------------- KV Cache V2 Scheduler IT (multi-GPU) ---------------
-  - kv_cache/test_kv_cache_v2_scheduler.py::TestKVCacheV2DSv3Lite::test_mtp_draft_tokens
-  - kv_cache/test_kv_cache_v2_scheduler.py::TestKVCacheV2DSv3Lite::test_mtp_chunked_draft_tokens
-  - kv_cache/test_kv_cache_v2_scheduler.py::TestKVCacheV2DSv3Lite::test_mtp_eviction
   # ------------- NVBug 6025177: trtllm-serve cross-request KV contamination (OpenAI) ---------------
   - test_e2e.py::test_openai_kv_cache_contamination TIMEOUT (120)
 - condition:
@@ -81,7 +99,6 @@ l0_dgx_b200:
   - unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_multi_gpu -k "DEEPGEMM and not MEGAMOE_DEEPGEMM"
   # --- MEGAMOE_DEEPGEMM (W4A8_MXFP4_MXFP8 only) ---
   - unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_multi_gpu -k "MEGAMOE_DEEPGEMM"
-  - unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_single_gpu -k "MEGAMOE_DEEPGEMM"
   # ------------- MoE: test_multi_gpu_eplb ---------------
   - unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_multi_gpu_eplb
 - condition:
@@ -165,8 +182,6 @@ l0_dgx_b200:
   - 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
-  - unittest/_torch/visual_gen/test_flux_pipeline.py::TestFluxParallelism::test_ulysses_2gpu_correctness
-  - unittest/_torch/visual_gen/test_flux_pipeline.py::TestFluxCombinedOptimizations::test_all_optimizations_combined
 - condition:
     ranges:
       system_gpu_count:
@@ -192,7 +207,6 @@ l0_dgx_b200:
   - accuracy/test_llm_api_pytorch.py::TestDeepSeekV32::test_fp8_blockscale[baseline_fp8kv] TIMEOUT (60)
   - accuracy/test_llm_api_pytorch.py::TestDeepSeekV32::test_fp8_blockscale[latency] TIMEOUT (60)
   - accuracy/test_llm_api_pytorch.py::TestDeepSeekV32::test_fp8_blockscale[disable_skip_indexer] TIMEOUT (60)
-  - accuracy/test_llm_api_pytorch.py::TestDeepSeekV32::test_nvfp4_attn_multi_gpus TIMEOUT (60)
   - accuracy/test_llm_api_pytorch.py::TestDeepSeekV32::test_nvfp4_multi_gpus[baseline_fp8kv] TIMEOUT (60)
   - accuracy/test_llm_api_pytorch.py::TestDeepSeekV32::test_nvfp4_multi_gpus[latency] TIMEOUT (60)
   - accuracy/test_llm_api_pytorch.py::TestDeepSeekV32::test_nvfp4_multi_gpus[disable_skip_indexer] TIMEOUT (60)
@@ -305,30 +319,22 @@ l0_dgx_b200:
   - accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_nvfp4_4gpus[moe_backend=CUTEDSL-mtp_nextn=2-ep4-fp8kv=False-attention_dp=True-cuda_graph=False-overlap_scheduler=False-low_precision_combine=True-torch_compile=False]
   - accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_nvfp4_4gpus[moe_backend=CUTEDSL-mtp_nextn=2-ep4-fp8kv=True-attention_dp=True-cuda_graph=True-overlap_scheduler=True-low_precision_combine=True-torch_compile=False]
   - accuracy/test_llm_api_pytorch.py::TestLlama3_3_70BInstruct::test_fp4_tp2pp2[torch_compile=False-enable_gemm_allreduce_fusion=False]
-  - examples/test_visual_gen.py::test_visual_gen_quickstart
-  - examples/test_visual_gen.py::test_visual_gen_api_walkthrough
   - examples/test_visual_gen.py::test_wan_t2v_example
-  - examples/test_visual_gen.py::test_flux1_lpips_against_golden
-  - examples/test_visual_gen.py::test_flux2_lpips_against_golden
-  - examples/test_visual_gen.py::test_ltx2_lpips_against_golden
-  - examples/test_visual_gen.py::test_wan21_t2v_lpips_against_golden
-  - examples/test_visual_gen.py::test_wan22_t2v_lpips_against_golden
   - examples/test_visual_gen_multi_gpu.py::test_wan22_t2v_lpips_against_golden_multi_gpu[ulysses4]
   - examples/test_visual_gen_multi_gpu.py::test_wan22_t2v_lpips_against_golden_multi_gpu[cfg2_ulysses2]
   - examples/test_visual_gen_multi_gpu.py::test_wan22_t2v_lpips_against_golden_multi_gpu[ulysses2_ring2]
   - examples/test_visual_gen_multi_gpu.py::test_wan22_t2v_lpips_against_golden_multi_gpu[attn2d_2x2]
   - examples/test_visual_gen.py::test_vbench_dimension_score_wan
   - examples/test_visual_gen.py::test_vbench_dimension_score_wan22_a14b_fp8
   - examples/test_visual_gen.py::test_vbench_dimension_score_wan22_a14b_nvfp4
-  - visual_gen/test_visual_gen_benchmark.py::test_offline_benchmark
-  - visual_gen/test_visual_gen_benchmark.py::test_online_benchmark[openai-videos]
   - examples/test_visual_gen.py::test_vbench_dimension_score_ltx2_bf16
   - examples/test_visual_gen.py::test_vbench_dimension_score_ltx2_fp8
   - accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B_Instruct_2507::test_skip_softmax_attention_4gpus[target_sparsity_0.5-fp8kv=False]
   - accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B_Instruct_2507::test_skip_softmax_attention_4gpus[target_sparsity_0.5-fp8kv=True]
   - accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B_Instruct_2507::test_skip_softmax_attention_4gpus[target_sparsity_0.9-fp8kv=False]
   - accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B_Instruct_2507::test_skip_softmax_attention_4gpus[target_sparsity_0.9-fp8kv=True]
   - disaggregated/test_disaggregated.py::test_disaggregated_mamba_conc_greater_than_mbs[NVIDIA-Nemotron-3-Super-120B-A12B-FP8]
+  - accuracy/test_llm_api_pytorch.py::TestDeepSeekV32::test_nvfp4_attn_multi_gpus TIMEOUT (60)
 # ------------- AutoDeploy Backend Stages ---------------
 - condition:
     ranges: