You are not logged in.
Has anyone tried Google's amazing congestion control algorithm, BBR, introduced in Linux 4.9?
https://lwn.net/Articles/701165 says these bits of code "allow network protocols (usually TCP) to maximize the throughput of any given connection while simultaneously sharing the available bandwidth equitably with other users."
Here are few benchmarks:
https://github.com/AOSC-Dev/aosc-os-abbs/issues/523
however, the result covers only bbr/fq_codel, not bbr/fq.
I also found another benchmark result:
Hybla: 689Kbps
BBR: 19.1Mbps
Offline
Mod note: Moving thread to GNU/Linux discussion.
Offline
I have been using CONFIG_DEFAULT_TCP_CONG="bbr" since just after this post. Honestly, I cannot tell the difference between BBR and Reno for the workload I push through the pipe...
Arch Linux + sway
Debian Testing + GNOME/sway
NetBSD 64-bit + Xfce
Offline