Support page size > 1 (#4356)

test/srt/test_mla_deepseek_v3.py

@@ -13,7 +13,7 @@ from sglang.test.test_utils import (
 )
 
 
-class TestDeepseekV3(unittest.TestCase):
+class TestMLADeepseekV3(unittest.TestCase):
     @classmethod
     def setUpClass(cls):
         cls.model = "lmsys/sglang-ci-dsv3-test"

test/srt/test_page_size.py

+import os
+import unittest
+from types import SimpleNamespace
+
+from sglang.srt.utils import kill_process_tree
+from sglang.test.run_eval import run_eval
+from sglang.test.test_utils import (
+    DEFAULT_MODEL_NAME_FOR_TEST,
+    DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
+    DEFAULT_URL_FOR_TEST,
+    popen_launch_server,
+)
+
+
+class TestPageSize(unittest.TestCase):
+    @classmethod
+    def setUpClass(cls):
+        os.environ["SGLANG_DEBUG_MEMORY_POOL"] = "1"
+        cls.model = DEFAULT_MODEL_NAME_FOR_TEST
+        cls.base_url = DEFAULT_URL_FOR_TEST
+        cls.process = popen_launch_server(
+            cls.model,
+            cls.base_url,
+            timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
+            other_args=["--page-size", 4, "--chunked-prefill-size", 128],
+        )
+
+    @classmethod
+    def tearDownClass(cls):
+        kill_process_tree(cls.process.pid)
+
+    def test_mmlu(self):
+        args = SimpleNamespace(
+            base_url=self.base_url,
+            model=self.model,
+            eval_name="mmlu",
+            num_examples=64,
+            num_threads=32,
+        )
+
+        metrics = run_eval(args)
+        self.assertGreaterEqual(metrics["score"], 0.65)
+
+
+if __name__ == "__main__":
+    unittest.main()

test/srt/test_retract_decode.py

+import os
 import unittest
 from types import SimpleNamespace
 
@@ -14,6 +15,8 @@ from sglang.test.test_utils import (
 class TestRetractDecode(unittest.TestCase):
     @classmethod
     def setUpClass(cls):
+        os.environ["SGLANG_TEST_RETRACT"] = "1"
+
         cls.model = DEFAULT_MODEL_NAME_FOR_TEST
         cls.base_url = DEFAULT_URL_FOR_TEST
         cls.process = popen_launch_server(
@@ -37,5 +40,20 @@ class TestRetractDecode(unittest.TestCase):
         self.assertGreaterEqual(metrics["score"], 0.65)
 
 
+class TestRetractDecodeChunkCache(unittest.TestCase):
+    @classmethod
+    def setUpClass(cls):
+        os.environ["SGLANG_TEST_RETRACT"] = "1"
+
+        cls.model = DEFAULT_MODEL_NAME_FOR_TEST
+        cls.base_url = DEFAULT_URL_FOR_TEST
+        cls.process = popen_launch_server(
+            cls.model,
+            cls.base_url,
+            timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
+            other_args=["--disable-radix-cache", "--chunked-prefill-size", 128],
+        )
+
+
 if __name__ == "__main__":
     unittest.main()