[None][ci] add DGX_H100-2_GPUs-PyTorch-Others-1 pipeline (#7629)

Signed-off-by: junq <[email protected]>

Author: QiJune

jenkins/L0_Test.groovy

@@ -1967,6 +1967,7 @@ def launchTestJobs(pipeline, testFilter)
     x86TestConfigs = [
         "DGX_H100-4_GPUs-PyTorch-DeepSeek-1": ["dgx-h100-x4", "l0_dgx_h100", 1, 2, 4],
         "DGX_H100-4_GPUs-PyTorch-DeepSeek-2": ["dgx-h100-x4", "l0_dgx_h100", 2, 2, 4],
+        "DGX_H100-2_GPUs-PyTorch-Others-1": ["dgx-h100-x2", "l0_dgx_h100", 1, 1, 2],
         "DGX_H100-4_GPUs-PyTorch-Others-1": ["dgx-h100-x4", "l0_dgx_h100", 1, 1, 4],
         "DGX_H100-4_GPUs-CPP-1": ["dgx-h100-x4", "l0_dgx_h100", 1, 1, 4],
         "A10-PyTorch-1": ["a10", "l0_a10", 1, 1],
@@ -2353,9 +2354,9 @@ def launchTestJobs(pipeline, testFilter)
         }, {}, true)
     }]}
 
-    multiGpuJobs = parallelJobs.findAll{(it.key.contains("4_GPUs") || it.key.contains("8_GPUs")) && !it.key.contains("Post-Merge")}
+    multiGpuJobs = parallelJobs.findAll{(it.key.contains("2_GPUs") || it.key.contains("4_GPUs") || it.key.contains("8_GPUs")) && !it.key.contains("Post-Merge")}
     println multiGpuJobs.keySet()
-    multiGpuJobsPostMerge = parallelJobs.findAll{(it.key.contains("4_GPUs") || it.key.contains("8_GPUs")) && it.key.contains("Post-Merge")}
+    multiGpuJobsPostMerge = parallelJobs.findAll{(it.key.contains("2_GPUs") || it.key.contains("4_GPUs") || it.key.contains("8_GPUs")) && it.key.contains("Post-Merge")}
 
     parallelJobs += docBuildJobs
     parallelJobs += sanityCheckJobs

tests/integration/test_lists/test-db/l0_dgx_h100.yml

@@ -1,5 +1,40 @@
 version: 0.0.1
 l0_dgx_h100:
+- condition:
+    ranges:
+      system_gpu_count:
+        gte: 2
+        lte: 2
+    wildcards:
+      gpu:
+      - '*h100*'
+      linux_distribution_name: ubuntu*
+    terms:
+      stage: pre_merge
+      backend: pytorch
+      auto_trigger: others
+  tests:
+  - unittest/llmapi/test_llm_multi_gpu_pytorch.py -m "gpu2"
+  - unittest/_torch/multi_gpu -m "not post_merge" TIMEOUT (90)
+  - unittest/_torch/auto_deploy/unit/multigpu
+  - unittest/_torch/modeling/test_modeling_pixtral.py::test_tensor_parallelism
+  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_eagle3[eagle3_one_model=False-overlap_scheduler=False]
+  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_eagle3[eagle3_one_model=True-overlap_scheduler=True]
+  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_guided_decoding[xgrammar]
+  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_guided_decoding_with_eagle3[xgrammar-eagle3_one_model=True]
+  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_guided_decoding_with_eagle3[xgrammar-eagle3_one_model=False]
+  - accuracy/test_disaggregated_serving.py::TestQwen3_8B::test_auto_dtype[False]
+  - accuracy/test_disaggregated_serving.py::TestQwen3_8B::test_auto_dtype[True]
+  - accuracy/test_disaggregated_serving.py::TestQwen3_8B::test_chunked_prefill
+  - accuracy/test_disaggregated_serving.py::TestQwen3_8B::test_nixl_backend
+  - accuracy/test_disaggregated_serving.py::TestDeepSeekV3Lite::test_nixl_backend
+  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_ngram
+  - accuracy/test_disaggregated_serving.py::TestGemma3_1BInstruct::test_auto_dtype[False]
+  - accuracy/test_disaggregated_serving.py::TestGemma3_1BInstruct::test_auto_dtype[True]
+  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_auto_dtype[False]
+  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_auto_dtype[True]
+  # ------------- AutoDeploy tests ---------------
+  - accuracy/test_llm_api_autodeploy.py::TestLlama3_1_8B::test_auto_dtype
 - condition:
     ranges:
       system_gpu_count:
@@ -15,9 +50,7 @@ l0_dgx_h100:
       auto_trigger: others
   tests:
   # ------------- PyTorch tests ---------------
-  - unittest/_torch/multi_gpu -m "not post_merge" TIMEOUT (90)
-  - unittest/_torch/auto_deploy/unit/multigpu
-  - unittest/llmapi/test_llm_multi_gpu_pytorch.py -m "gpu4 or gpu2"
+  - unittest/llmapi/test_llm_multi_gpu_pytorch.py -m "gpu4"
   - accuracy/test_llm_api_pytorch.py::TestLlama3_1_8BInstruct::test_bfloat16_4gpus[tp4-attn_backend=TRTLLM-torch_compile=False]
   - accuracy/test_llm_api_pytorch.py::TestLlama3_1_8BInstruct::test_bfloat16_4gpus[tp2pp2-attn_backend=TRTLLM-torch_compile=False]
   - accuracy/test_llm_api_pytorch.py::TestLlama3_1_8BInstruct::test_bfloat16_4gpus[tp2pp2-attn_backend=TRTLLM-torch_compile=True]
@@ -35,19 +68,6 @@ l0_dgx_h100:
   - disaggregated/test_disaggregated.py::test_disaggregated_ctxpp2_gentp2[TinyLlama-1.1B-Chat-v1.0]
   - disaggregated/test_disaggregated.py::test_disaggregated_ctxpp4_gentp4[TinyLlama-1.1B-Chat-v1.0]
   - disaggregated/test_disaggregated.py::test_disaggregated_genbs1[TinyLlama-1.1B-Chat-v1.0]
-  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_auto_dtype[False]
-  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_auto_dtype[True]
-  - accuracy/test_disaggregated_serving.py::TestGemma3_1BInstruct::test_auto_dtype[False]
-  - accuracy/test_disaggregated_serving.py::TestGemma3_1BInstruct::test_auto_dtype[True]
-  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_ngram
-  - accuracy/test_disaggregated_serving.py::TestQwen3_8B::test_auto_dtype[False]
-  - accuracy/test_disaggregated_serving.py::TestQwen3_8B::test_auto_dtype[True]
-  - accuracy/test_disaggregated_serving.py::TestQwen3_8B::test_chunked_prefill
-  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_eagle3[eagle3_one_model=False-overlap_scheduler=False]
-  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_eagle3[eagle3_one_model=True-overlap_scheduler=True]
-  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_guided_decoding[xgrammar]
-  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_guided_decoding_with_eagle3[xgrammar-eagle3_one_model=True]
-  - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_guided_decoding_with_eagle3[xgrammar-eagle3_one_model=False]
   - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_tp_pp_symmetric[GSM8K-tp1pp2]
   - 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[GSM8K-tp2pp1]
@@ -58,13 +78,8 @@ l0_dgx_h100:
   - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_ctx_pp_gen_tp_asymmetric[MMLU-gen_tp=2-ctx_pp=2]
   - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_multi_instance[GSM8K]
   - accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_multi_instance[MMLU]
-  - accuracy/test_disaggregated_serving.py::TestQwen3_8B::test_nixl_backend
-  - accuracy/test_disaggregated_serving.py::TestDeepSeekV3Lite::test_nixl_backend
   - test_e2e.py::test_ptp_quickstart_advanced_bs1
   - test_e2e.py::test_ptp_quickstart_advanced_deepseek_v3_lite_4gpus_adp_balance[DeepSeek-V3-Lite-FP8-DeepSeek-V3-Lite/fp8]
-  - unittest/_torch/modeling/test_modeling_pixtral.py::test_tensor_parallelism
-  # ------------- AutoDeploy tests ---------------
-  - accuracy/test_llm_api_autodeploy.py::TestLlama3_1_8B::test_auto_dtype
 - condition:
     ranges:
       system_gpu_count:

tests/unittest/_torch/multi_gpu/test_lowprecision_allreduce.py

@@ -249,9 +249,9 @@ def test(self, mode="acc"):
     ids=lambda x: f"size{x}")
 @pytest.mark.parametrize(
     "mpi_pool_executor",
-    [2, 4],  # 8
-    ids=["tp_size_2", "tp_size_4"],
-    indirect=True)  # "tp_size_8"
+    [2],  # 4, 8
+    ids=["tp_size_2"],
+    indirect=True)  # "tp_size_4", "tp_size_8"
 def test_lowprecision_allreduce_acc(dtype, strategy, message_size,
                                     mpi_pool_executor):
     """