test: [nvbug: 5234494] skip_pre_ada for fp8 cases (#3718)

* skip_pre_ada for fp8 cases

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

* update

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

* update after rebase

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

---------

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

Author: crazydemo

tests/integration/defs/examples/test_llama.py

@@ -4669,6 +4669,7 @@ def test_llm_llama_lookahead_xqa_fp8_1gpu(llama_example_root, llama_model_root,
     venv_check_call(llm_venv, summary_cmd)
 
 
+@skip_pre_ada
 @pytest.mark.skip_less_device_memory(80000)
 @pytest.mark.parametrize("code_llama_model_root", ['CodeLlama-7b-Instruct'],
                          indirect=True)

tests/integration/defs/examples/test_mistral.py

@@ -355,6 +355,7 @@ def test_llm_mistral_v1_smooth_quant_4gpus(llama_example_root,
                         summary_cmd)
 
 
+@skip_pre_ada
 @pytest.mark.parametrize("run_type", ['inference', 'summarization'])
 @pytest.mark.parametrize("mistral_nemo_model_root", ['Mistral-Nemo-12b-Base'],
                          indirect=True)

tests/integration/defs/test_e2e.py

@@ -1605,9 +1605,11 @@ def test_ptp_quickstart_advanced_deepseek_r1_8gpus(llm_root, llm_venv,
     pytest.param('Mixtral-8x7B-NVFP4',
                  'nvfp4-quantized/Mixtral-8x7B-Instruct-v0.1',
                  marks=skip_pre_blackwell),
-    pytest.param('Nemotron-Ultra-253B',
-                 'nemotron-nas/Llama-3_1-Nemotron-Ultra-253B-v1',
-                 marks=skip_pre_hopper),
+    pytest.param(
+        'Nemotron-Ultra-253B',
+        'nemotron-nas/Llama-3_1-Nemotron-Ultra-253B-v1',
+        marks=[skip_pre_hopper,
+               pytest.mark.skip_less_device_memory(140000)]),
 ])
 def test_ptp_quickstart_advanced_8gpus(llm_root, llm_venv, model_name,
                                        model_path):