Managing CPU and network resources in the browser for large video grids

Agora has been researching the best ways to manage large numbers of remote video streams and support grids of up to 64 people (8×8) by detecting CPU and network resource limits from within the web browser environment. This is a difficult problem because CPU information is not available to Javascript and network stats can make it hard to distinguish between congestion and transmission error packet loss. In this session, I will look at the available WebRTC stats, some newly derived stats, and how to adapt in real time to both CPU and network constraints of the local device.
Speaker:
Ben Weekes, Senior Architect, Agora
Thank you to our sponsors!
Google - webrtc.org
Agora - www.agora.io
Daily - www.daily.co/
Dolby - bit.ly/3vgy7qy
Element - ems.element.io/
Intel - bit.ly/31AtHRy
RingCentral - developers.ringcentral.com/
SUBSCRIBE TO OUR WEEKLY: webrtcweekly.com/
00:00 - Intro
01:26 - Requirements for large grids
03:36 - CPU estimation alternatives
08:56 - Examples
13:16 - Summary

[@pratimvlogs4177]

Great talk. Specially about the render rate volatility

[@catherinekim1860]

This is so informative!

[@AS-oo5wy]

Priceless information that is nowhere else to find.

[@ceddymuhoza]

Please share demo code for these apps