Hicache Storage Layer Prototype (#7704)

This commit is contained in:
Zhiqiang Xie
2025-07-18 00:20:19 -07:00
committed by GitHub
parent 7891bac16b
commit 9d33fcfb8e
9 changed files with 714 additions and 4 deletions

View File

@@ -64,6 +64,7 @@ suites = {
TestFile("test_fused_moe.py", 30),
TestFile("test_hicache.py", 116),
TestFile("test_hicache_mla.py", 127),
TestFile("test_hicache_storage.py", 127),
TestFile("test_hidden_states.py", 55),
TestFile("test_int8_kernel.py", 8),
TestFile("test_input_embeddings.py", 38),

View File

@@ -0,0 +1,55 @@
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,
CustomTestCase,
popen_launch_server,
)
class TestHiCache(CustomTestCase):
@classmethod
def setUpClass(cls):
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=[
"--enable-hierarchical-cache",
"--mem-fraction-static",
0.7,
"--hicache-size",
100,
"--page-size",
"64",
"--hicache-storage-backend",
"file",
],
)
@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()