10-10-2024, 12:42 AM
(10-09-2024, 10:16 PM)Zeable Wrote:(10-09-2024, 09:43 PM)Tim Curtis Wrote: Assuming the network is not contributing to the issue I'd try swapping to see if you can isolate the issue.
Swap in a different Pi as the receiver, a 3B+ for example
Swap to the 5GHz band
Will work on moving IPs and WiFi bands but wanted to correct my statement that all are sub 1ms. I did go back and retest and obviously not sub 1ms so wanted to share more accurate information.
Sender to receiver 1
50 packets transmitted, 50 received, 0% packet loss, time 428ms
rtt min/avg/max/mdev = 2.395/14.945/86.341/19.269 ms, pipe 9, ipg/ewma 8.741/30.349 ms
Sender to receiver 2
50 packets transmitted, 50 received, 0% packet loss, time 314ms
rtt min/avg/max/mdev = 2.081/7.073/37.599/7.105 ms, pipe 3, ipg/ewma 6.415/13.651 ms
Switch to sender
Success rate is 100 percent (100/100), round-trip min/avg/max = 2/5/40 ms
Switch to receiver 1
Success rate is 100 percent (100/100), round-trip min/avg/max = 2/5/54 ms
Switch to receiver 2
Success rate is 100 percent (100/100), round-trip min/avg/max = 2/5/22 ms
The sender to receiver ping stats show high avg rtt and really high standard deviation (mdev) which means there is a lot of variability in rtt. I'm guessing the Opus CODEC on the receivers can't cope with this and eventually just goes to white noise.
What I see on my network for example is avg rtt < 4.5ms and mdev < 1.5ms. It's a single Router with Sender connected via Ethernet and Receivers on 5GHz wireless.
