EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[CI] Reorganize the 8 gpu tests (#6192)

Browse Source
This commit is contained in:
Lianmin Zheng
2025-05-11 10:55:06 -07:00
committed by GitHub
parent 01bdbf7f80
commit 03227c5fa6
6 changed files with 15 additions and 82 deletions
Download Patch File Download Diff File Expand all files Collapse all files

52
test/srt/test_pp_single_node.py
View File

@@ -9,13 +9,10 @@ import time
import unittest
from types import SimpleNamespace
import requests
from sglang.bench_one_batch_server import BenchArgs as OneBatchBenchArgs
from sglang.srt.server_args import ServerArgs
from sglang.srt.utils import kill_process_tree
from sglang.test.few_shot_gsm8k import run_eval
from sglang.test.runners import DEFAULT_PROMPTS
from sglang.test.test_utils import (
DEFAULT_MODEL_NAME_FOR_TEST,
DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
@@ -28,17 +25,16 @@ from sglang.test.test_utils import (
class TestPPAccuracy(unittest.TestCase):
@classmethod
def setUpClass(cls):
# These config helps find a leak.
os.environ["SGLANG_IS_IN_CI"] = "1"
cls.base_url = "http://127.0.0.1:23333"
cls.process = popen_launch_server(
DEFAULT_MODEL_NAME_FOR_TEST,
cls.base_url,
timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
other_args=[
"--tp-size",
2,
"--pp-size",
4,
"--disable-overlap-schedule",
"--chunked-prefill-size",
256,
],
@@ -66,49 +62,6 @@ class TestPPAccuracy(unittest.TestCase):
time.sleep(5)
# class TestPPAccuracyFlashInfer(unittest.TestCase):
# @classmethod
# def setUpClass(cls):
# # These config helps find a leak.
# os.environ["SGLANG_IS_IN_CI"] = "1"
# cls.base_url = "http://127.0.0.1:23333"
# cls.process = popen_launch_server(
# DEFAULT_MODEL_NAME_FOR_TEST,
# cls.base_url,
# timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
# other_args=[
# "--pp-size",
# 4,
# "--disable-overlap-schedule",
# "--attention-backend",
# "flashinfer",
# "--chunked-prefill-size",
# 256,
# ],
# )
#
# @classmethod
# def tearDownClass(cls):
# kill_process_tree(cls.process.pid)
#
# def test_gsm8k(self):
# args = SimpleNamespace(
# num_shots=5,
# data_path=None,
# num_questions=200,
# max_new_tokens=512,
# parallel=128,
# host="http://127.0.0.1",
# port=int(self.base_url.split(":")[-1]),
# )
# metrics = run_eval(args)
# print(f"{metrics=}")
#
# self.assertGreater(metrics["accuracy"], 0.75)
# # Wait a little bit so that the memory check happens.
# time.sleep(5)
class TestFixedBugs(unittest.TestCase):
def test_chunked_prefill_with_small_bs(self):
model = DEFAULT_MODEL_NAME_FOR_TEST
@@ -124,7 +77,6 @@ class TestFixedBugs(unittest.TestCase):
2,
"--pp-size",
2,
"--disable-overlap-schedule",
"--chunked-prefill",
256,
"--max-running-requests",