From 005a17f2b2070e755f96734b403773354a0e7d6c Mon Sep 17 00:00:00 2001
From: Ivy Zhang <25222398+crazydemo@users.noreply.github.com>
Date: Mon, 19 Jan 2026 20:42:59 +0800
Subject: [PATCH 1/3] update case kv cache fraction to avoid oom

Signed-off-by: Ivy Zhang <25222398+crazydemo@users.noreply.github.com>
---
 tests/integration/defs/accuracy/test_llm_api_pytorch.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/integration/defs/accuracy/test_llm_api_pytorch.py b/tests/integration/defs/accuracy/test_llm_api_pytorch.py
index db6ef3c4e8e4..1c9f4cf52ebd 100644
--- a/tests/integration/defs/accuracy/test_llm_api_pytorch.py
+++ b/tests/integration/defs/accuracy/test_llm_api_pytorch.py
@@ -4993,7 +4993,7 @@ def test_fp8_8gpus(self):
             pytest.skip(f"Model directory {model_dir} does not exist")
 
         # Configure model settings
-        kv_cache_config = KvCacheConfig(free_gpu_memory_fraction=0.8,
+        kv_cache_config = KvCacheConfig(free_gpu_memory_fraction=0.4,
                                         enable_block_reuse=True,
                                         enable_partial_reuse=False,
                                         dtype="fp8")

From 1b76a9efb2381f76a116e9b2d05874c8b7f37a13 Mon Sep 17 00:00:00 2001
From: Ivy Zhang <25222398+crazydemo@users.noreply.github.com>
Date: Mon, 19 Jan 2026 20:53:02 +0800
Subject: [PATCH 2/3] move long bench case from stress test list to core list

Signed-off-by: Ivy Zhang <25222398+crazydemo@users.noreply.github.com>
---
 tests/integration/test_lists/qa/llm_function_core.txt   | 2 ++
 tests/integration/test_lists/qa/llm_function_stress.txt | 2 --
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/tests/integration/test_lists/qa/llm_function_core.txt b/tests/integration/test_lists/qa/llm_function_core.txt
index 1928574cf389..ef01d699d82b 100644
--- a/tests/integration/test_lists/qa/llm_function_core.txt
+++ b/tests/integration/test_lists/qa/llm_function_core.txt
@@ -117,6 +117,8 @@ accuracy/test_llm_api_pytorch.py::TestDeepSeekR1::test_fp8_blockscale[throughput
 accuracy/test_llm_api_pytorch.py::TestDeepSeekR1::test_fp8_blockscale[throughput_mtp_trtllm]
 accuracy/test_llm_api_pytorch.py::TestDeepSeekR1::test_fp8_blockscale_chunked_prefill[latency]
 accuracy/test_llm_api_pytorch.py::TestDeepSeekR1::test_fp8_blockscale_chunked_prefill[throughput]
+accuracy/test_llm_api_pytorch.py::TestDeepSeekR1LongBenchV2::test_fp8_8gpus TIMEOUT (240)
+accuracy/test_llm_api_pytorch.py::TestDeepSeekR1LongBenchV2::test_nvfp4_4gpus TIMEOUT (240)
 accuracy/test_llm_api_pytorch.py::TestDeepSeekV32::test_fp8_blockscale[baseline]
 accuracy/test_llm_api_pytorch.py::TestDeepSeekV32::test_fp8_blockscale[baseline_mtp1]
 accuracy/test_llm_api_pytorch.py::TestDeepSeekV32::test_fp8_blockscale[baseline_fp8kv]
diff --git a/tests/integration/test_lists/qa/llm_function_stress.txt b/tests/integration/test_lists/qa/llm_function_stress.txt
index a4f744b469d6..7b14358fd3ed 100644
--- a/tests/integration/test_lists/qa/llm_function_stress.txt
+++ b/tests/integration/test_lists/qa/llm_function_stress.txt
@@ -3,5 +3,3 @@ stress_test/stress_test.py::test_run_stress_test[DeepSeek-V3_tp8-stress_time_360
 stress_test/stress_test.py::test_run_stress_test[DeepSeek-R1_tp8-stress_time_3600s_timeout_5400s-MAX_UTILIZATION-pytorch-stress-test-with-accuracy]
 disaggregated/test_disaggregated.py::test_disaggregated_stress_test[input8k-output1k-conc512-deepseek_r1_v2_fp4_stress]
 disaggregated/test_disaggregated.py::test_disaggregated_stress_test[input8k-output1k-conc512-gpt_oss_120b_stress]
-accuracy/test_llm_api_pytorch.py::TestDeepSeekR1LongBenchV2::test_fp8_8gpus
-accuracy/test_llm_api_pytorch.py::TestDeepSeekR1LongBenchV2::test_nvfp4_4gpus

From 1cd98158b3e6ac1357cc1997c37b0a2dc5fdfde7 Mon Sep 17 00:00:00 2001
From: Ivy Zhang <25222398+crazydemo@users.noreply.github.com>
Date: Tue, 20 Jan 2026 13:28:24 +0800
Subject: [PATCH 3/3] update nemotron super constraint

Signed-off-by: Ivy Zhang <25222398+crazydemo@users.noreply.github.com>
---
 tests/integration/defs/accuracy/test_llm_api_pytorch.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/tests/integration/defs/accuracy/test_llm_api_pytorch.py b/tests/integration/defs/accuracy/test_llm_api_pytorch.py
index 1c9f4cf52ebd..3ac40922087a 100644
--- a/tests/integration/defs/accuracy/test_llm_api_pytorch.py
+++ b/tests/integration/defs/accuracy/test_llm_api_pytorch.py
@@ -5319,6 +5319,7 @@ class TestNemotronV3Super(LlmapiAccuracyTestHarness):
     EXTRA_EVALUATOR_KWARGS = dict(chat_template_kwargs=dict(
         enable_thinking=False))
 
+    @skip_pre_hopper
     @pytest.mark.skip_less_device_memory(64000)
     @pytest.mark.skip_less_mpi_world_size(4)
     @pytest.mark.parametrize(