Expected throughput of AIOQUIC #490
Replies: 2 comments 1 reply
-
On an M1 Max MacBook with Python 3.12.2 on loopback I see 2024-04-08 05:43:17,337 INFO client Response received for GET /foo : 41943040 bytes in 0.8 s (420.358 Mbps) That's megabits in the measurement, so ~ 52.5 megabytes/s. Without knowing just how you made your response or other details (aioquic version, OS, python version) it's hard to say more. |
Beta Was this translation helpful? Give feedback.
-
I am seeing the same slow performance. Is there anything we can do to make this faster on Windows? Using python 3.12.7, aioquic 1.2.0, on windows 11 business. I added this route to examples/demo.py
At best for larger files I get 9 Mbps when running the client and server in two different terminals on the same machine. I don't see any unreasonable cpu or memory or disk load while it runs so I do not believe it is resource constrained. Dell Inspiron 16 Plus 7640
|
Beta Was this translation helpful? Give feedback.
-
What is the expected throughput of aioquic?
Not sure if I am doing something wrong here. I took the http3 example from the examples and modified the server to return to a post request a really large message (40 MB). Both the client and server are ran on my local machine, so there is no packet loss. I would expect such a transfer to take less than a second (as a baseline I used the streams native to asyncio). Yet, the throughput was barely 1 MB/s.
Am I missing something about this library? Or is this the intended throughput?
Beta Was this translation helpful? Give feedback.
All reactions