From e3a95077bc9aeb0fd0b939a3bbe48cf6c0d7083d Mon Sep 17 00:00:00 2001
From: Qiaolin Yu <liin1211@outlook.com>
Date: Thu, 15 Jan 2026 16:13:35 -0800
Subject: [PATCH] Add dpsk-r1-fp4 in nightly perf ci (#16882)

---
 .github/workflows/nightly-test-nvidia.yml     |  4 +-
 .../perf/test_dpsk_r1_fp4_4gpu_perf.py        | 74 +++++++++++++++++++
 2 files changed, 76 insertions(+), 2 deletions(-)
 create mode 100644 test/registered/perf/test_dpsk_r1_fp4_4gpu_perf.py

diff --git a/.github/workflows/nightly-test-nvidia.yml b/.github/workflows/nightly-test-nvidia.yml
index 8190c7e1d..c97c420fc 100644
--- a/.github/workflows/nightly-test-nvidia.yml
+++ b/.github/workflows/nightly-test-nvidia.yml
@@ -374,10 +374,10 @@ jobs:
           IS_BLACKWELL=1 bash scripts/ci/ci_install_dependency.sh
 
       - name: Run test
-        timeout-minutes: 60
+        timeout-minutes: 300
         run: |
           cd test
-          python3 run_suite.py --hw cuda --suite nightly-4-gpu-b200 --nightly --continue-on-error
+          python3 run_suite.py --hw cuda --suite nightly-4-gpu-b200 --nightly --continue-on-error --timeout-per-file 12000
 
   # Specialized B200 tests - 8 GPU, for specific backends and configs
   nightly-test-specialized-8-gpu-b200:
diff --git a/test/registered/perf/test_dpsk_r1_fp4_4gpu_perf.py b/test/registered/perf/test_dpsk_r1_fp4_4gpu_perf.py
new file mode 100644
index 000000000..b03c34337
--- /dev/null
+++ b/test/registered/perf/test_dpsk_r1_fp4_4gpu_perf.py
@@ -0,0 +1,74 @@
+import unittest
+
+from sglang.test.accuracy_test_runner import AccuracyTestParams
+from sglang.test.ci.ci_register import register_cuda_ci
+from sglang.test.performance_test_runner import PerformanceTestParams
+from sglang.test.run_combined_tests import run_combined_tests
+from sglang.test.test_utils import ModelLaunchSettings
+
+# Runs on B200 via nightly-4-gpu-b200 suite
+register_cuda_ci(est_time=2000, suite="nightly-4-gpu-b200", nightly=True)
+
+DEEPSEEK_R1_FP4_MODEL_PATH = "nvidia/DeepSeek-R1-0528-NVFP4-v2"
+
+
+class TestDeepseekR1FP4Unified(unittest.TestCase):
+    """Unified test class for DeepSeek-R1-0528-NVFP4-v2 performance and accuracy.
+
+    Two variants:
+    - basic: Standard TP=4
+    - mtp: TP=4 + EAGLE speculative decoding
+
+    Each variant runs BOTH:
+    - Performance test (using NightlyBenchmarkRunner)
+    - Accuracy test (using run_eval with mgsm_en)
+    """
+
+    def test_deepseek_r1_fp4_all_variants(self):
+        """Run performance and accuracy for all DeepSeek-R1-0528-NVFP4-v2 variants."""
+        # Define base arguments shared by most variants
+        base_args = [
+            "--tp=4",
+            "--trust-remote-code",
+            "--model-loader-extra-config",
+            '{"enable_multithread_load": true}',
+        ]
+        mtp_args = [
+            "--speculative-algorithm=EAGLE",
+            "--speculative-num-steps=3",
+            "--speculative-eagle-topk=1",
+            "--speculative-num-draft-tokens=4",
+            "--mem-frac=0.7",
+        ]
+
+        variants = [
+            # Variant: "basic" - Standard TP=4
+            ModelLaunchSettings(
+                DEEPSEEK_R1_FP4_MODEL_PATH,
+                tp_size=4,
+                extra_args=base_args,
+                variant="TP4",
+            ),
+            # Variant: "mtp" - TP=4 + EAGLE speculative decoding
+            ModelLaunchSettings(
+                DEEPSEEK_R1_FP4_MODEL_PATH,
+                tp_size=4,
+                extra_args=base_args + mtp_args,
+                variant="TP4+MTP",
+            ),
+        ]
+
+        run_combined_tests(
+            models=variants,
+            test_name="DeepSeek-R1-0528-NVFP4-v2 Unified",
+            accuracy_params=AccuracyTestParams(
+                dataset="gsm8k", baseline_accuracy=0.935
+            ),
+            performance_params=PerformanceTestParams(
+                profile_dir="performance_profiles_deepseek_r1_fp4",
+            ),
+        )
+
+
+if __name__ == "__main__":
+    unittest.main()