[https://nvbugs/5997534][fix] Skip Eagle3 AD test on pre-Hopper GPUs and unwaive

Eagle3 acceptance rate threshold (25%) is calibrated for Hopper hardware
where the draft model was trained. On L20 (Ada, SM 89) the acceptance
rate is ~22%, which is functional but below the threshold. Skip the test
on pre-Hopper GPUs and remove the global waive so it resumes running on
Hopper+.

Signed-off-by: Govind Ramnarayan <105831528+govind-ramnarayan@users.noreply.github.com>

Author: govind-ramnarayan

tests/integration/defs/accuracy/test_llm_api_autodeploy.py

@@ -18,7 +18,8 @@
 import pytest
 import torch
 import yaml
-from defs.conftest import get_llm_root, get_sm_version, skip_pre_blackwell
+from defs.conftest import (get_llm_root, get_sm_version, skip_pre_blackwell,
+                           skip_pre_hopper)
 from test_common.llm_data import hf_id_to_local_model_dir, llm_models_root
 
 from tensorrt_llm._torch.auto_deploy import LLM as AutoDeployLLM
@@ -317,6 +318,7 @@ def check_acceptance_rate(self, llm, min_acceptance_rate: float):
         """
         _check_acceptance_rate_stats(llm.get_stats(), min_acceptance_rate)
 
+    @skip_pre_hopper
     @pytest.mark.skip_less_device_memory(32000)
     def test_eagle3_one_model(self):
         """Test Eagle3 one-model speculative decoding accuracy on GSM8K."""

tests/integration/test_lists/waives.txt

@@ -324,7 +324,6 @@ accuracy/test_llm_api_pytorch.py::TestDeepSeekV32::test_nvfp4_multi_gpus_piecewi
 accuracy/test_llm_api_pytorch.py::TestGemma3_1BInstruct::test_auto_dtype SKIP (https://nvbugs/5992113)
 accuracy/test_llm_api_pytorch.py::TestLlama3_1_8BInstruct::test_chunked_prefill[use_temperature=False-attn_backend=TRTLLM] SKIP (https://nvbugs/5997547)
 disaggregated/test_disaggregated.py::test_disaggregated_overlap_transceiver_runtime_python[TinyLlama-1.1B-Chat-v1.0] SKIP (https://nvbugs/5997543)
-accuracy/test_llm_api_autodeploy.py::TestLlama3_1_8B_Instruct_Eagle3::test_eagle3_one_model SKIP (https://nvbugs/5997534)
 accuracy/test_llm_api_autodeploy.py::TestNemotronNanoV3::test_accuracy[fp8-4-trtllm] SKIP (https://nvbugs/5997046)
 accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_bfloat16_4gpus_python_scheduler[ep4-mtp_nextn=0] SKIP (https://nvbugs/5997051)
 perf/test_perf_sanity.py::test_e2e[aggr_upload-deepseek_v32_fp4_blackwell-v32_fp4_tep8_mtp3_8k1k] SKIP (https://nvbugs/5997092)