add ds v3 bf16 chunked_prefill on hopper

Signed-off-by: Ivy Zhang <[email protected]>

Author: crazydemo

tests/integration/defs/accuracy/test_llm_api_pytorch.py

@@ -1132,7 +1132,7 @@ class TestDeepSeekV3Lite(LlmapiAccuracyTestHarness):
     # Chunked Prefill for MLA can only be enabled on SM100
     @parametrize_with_ids(
         "enable_chunked_prefill",
-        [False, pytest.param(True, marks=skip_pre_blackwell)])
+        [False, pytest.param(True, marks=skip_pre_hopper)])
     @parametrize_with_ids("torch_compile", [False, True])
     @parametrize_with_ids("attention_dp,cuda_graph,overlap_scheduler",
                           [(False, False, False), (True, False, False),

tests/integration/test_lists/qa/llm_function_full.txt

@@ -510,6 +510,7 @@ accuracy/test_llm_api_pytorch.py::TestLlama4ScoutInstruct::test_fp4_chunked_pref
 accuracy/test_llm_api_pytorch.py::TestMixtral8x7B::test_fp8_tp2
 accuracy/test_llm_api_pytorch.py::TestMixtral8x7B::test_nvfp4_tp2
 accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_bfloat16[mtp_nextn=0-attention_dp=False-cuda_graph=False-overlap_scheduler=False-torch_compile=False-enable_chunked_prefill=False]
+accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_bfloat16[mtp_nextn=2-attention_dp=True-cuda_graph=True-overlap_scheduler=True-torch_compile=False-enable_chunked_prefill=True]
 accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_fp8_block_scales[mtp=disable-fp8kv=False-attention_dp=False-cuda_graph=False-overlap_scheduler=False-torch_compile=False]
 accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_nvfp4[moe_backend=CUTLASS-mtp_nextn=0-fp8kv=False-attention_dp=False-cuda_graph=False-overlap_scheduler=False-torch_compile=False]
 accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_nvfp4[moe_backend=CUTLASS-mtp_nextn=2-fp8kv=False-attention_dp=False-cuda_graph=False-overlap_scheduler=False-torch_compile=False]