fix: enable multi-threading for h200 tests (#16413)

test/srt/ep/test_deepep_large.py

@@ -174,6 +174,8 @@ class TestDeepseekV32TBO(CustomTestCase):
             "deepep",
             "--cuda-graph-max-bs",
             "256",
+            "--model-loader-extra-config",
+            '{"enable_multithread_load": true, "num_threads": 64}',
         ]
         cls.process = popen_launch_server(
             cls.model,

test/srt/test_deepseek_v32_basic.py

@@ -28,6 +28,8 @@ class TestDeepseekV32DP(CustomTestCase):
             "--dp",
             "8",
             "--enable-dp-attention",
+            "--model-loader-extra-config",
+            '{"enable_multithread_load": true, "num_threads": 64}',
         ]
         cls.process = popen_launch_server(
             cls.model,
@@ -83,6 +85,8 @@ class TestDeepseekV32TP(CustomTestCase):
             "--trust-remote-code",
             "--tp",
             "8",
+            "--model-loader-extra-config",
+            '{"enable_multithread_load": true, "num_threads": 64}',
         ]
         cls.process = popen_launch_server(
             cls.model,

test/srt/test_deepseek_v32_cp_single_node.py

@@ -52,6 +52,8 @@ class TestDeepseekV32CP(CustomTestCase):
             "--enable-nsa-prefill-context-parallel",
             "--nsa-prefill-cp-mode",
             "in-seq-split",
+            "--model-loader-extra-config",
+            '{"enable_multithread_load": true, "num_threads": 64}',
         ]
         cls.process = popen_launch_server(
             cls.model,
@@ -123,6 +125,8 @@ class TestDeepseekV32CPMode1(CustomTestCase):
             "--enable-nsa-prefill-context-parallel",
             "--nsa-prefill-cp-mode",
             "round-robin-split",
+            "--model-loader-extra-config",
+            '{"enable_multithread_load": true, "num_threads": 64}',
         ]
         cls.process = popen_launch_server(
             cls.model,

test/srt/test_deepseek_v32_mtp.py

@@ -40,6 +40,8 @@ class TestDeepseekV32DPMTP(CustomTestCase):
             "4",
             "--mem-frac",
             "0.7",
+            "--model-loader-extra-config",
+            '{"enable_multithread_load": true, "num_threads": 64}',
         ]
         cls.process = popen_launch_server(
             cls.model,
@@ -120,6 +122,8 @@ class TestDeepseekV32TPMTP(CustomTestCase):
             "4",
             "--mem-frac",
             "0.7",
+            "--model-loader-extra-config",
+            '{"enable_multithread_load": true, "num_threads": 64}',
         ]
         cls.process = popen_launch_server(
             cls.model,

test/srt/test_deepseek_v3_basic.py

@@ -22,7 +22,13 @@ class TestDeepseekV3Basic(CustomTestCase):
     def setUpClass(cls):
         cls.model = FULL_DEEPSEEK_V3_MODEL_PATH
         cls.base_url = DEFAULT_URL_FOR_TEST
-        other_args = ["--trust-remote-code", "--tp", "8"]
+        other_args = [
+            "--trust-remote-code",
+            "--tp",
+            "8",
+            "--model-loader-extra-config",
+            '{"enable_multithread_load": true, "num_threads": 64}',
+        ]
         cls.process = popen_launch_server(
             cls.model,
             cls.base_url,

test/srt/test_deepseek_v3_mtp.py

@@ -36,6 +36,8 @@ class TestDeepseekV3MTP(CustomTestCase):
             "1",
             "--speculative-num-draft-tokens",
             "4",
+            "--model-loader-extra-config",
+            '{"enable_multithread_load": true, "num_threads": 64}',
         ]
         if not is_in_amd_ci():
             other_args += ["--mem-frac", "0.7"]