xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-04-06 09:47:02 +08:00
Code Issues Packages Projects Releases Wiki Activity

run format

Browse Source 
Signed-off-by: clark <panf2333@gmail.com>
This commit is contained in:
clark 2025-01-16 20:13:51 +08:00
parent 16ed827378
commit 6c8fae82dd
2 changed files with 14 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
benchmarks/disagg_benchmarks/zmq/visualize_benchmark_results_zmq_http.py
View File

@ -5,7 +5,9 @@ import pandas as pd
if __name__ == "__main__":
data = []
for name in ['disagg_prefill_http', 'disagg_prefill_zmq', 'chunked_prefill']:
for name in [
'disagg_prefill_http', 'disagg_prefill_zmq', 'chunked_prefill'
]:
for qps in [2, 4, 6, 8, 10, 12]:
with open(f"results/{name}-qps-{qps}.json") as f:
x = json.load(f)
@ -18,7 +20,6 @@ if __name__ == "__main__":
dis_zmq_df = df[df['name'] == 'disagg_prefill_zmq']
chu_df = df[df['name'] == 'chunked_prefill']
plt.style.use('bmh')
plt.rcParams['font.size'] = 20
@ -51,7 +52,6 @@ if __name__ == "__main__":
fig.savefig(f'results/http_zmq_chunk/{key}.png')
plt.close(fig)
fig1, ax1 = plt.subplots(figsize=(11, 7))
plt.plot(dis_http_df['qps'],
dis_http_df[key],

23
vllm/entrypoints/disagg_connector.py
View File

@ -212,22 +212,21 @@ async def run_disagg_connector(args, **uvicorn_kwargs) -> None:
if __name__ == "__main__":
# NOTE(simon):
# NOTE(simon):
# This section should be in sync with vllm/scripts.py for CLI entrypoints.
parser = FlexibleArgumentParser(
description="vLLM disagg zmq server.")
parser = FlexibleArgumentParser(description="vLLM disagg zmq server.")
parser.add_argument("--port",
type=int,
default=8000,
help="The fastapi server port")
type=int,
default=8000,
help="The fastapi server port")
parser.add_argument("--prefill-addr",
type=str,
required=True,
help="The prefill address IP:PORT")
type=str,
required=True,
help="The prefill address IP:PORT")
parser.add_argument("--decode-addr",
type=str,
required=True,
help="The decode address IP:PORT")
type=str,
required=True,
help="The decode address IP:PORT")
args = parser.parse_args()