R7800 congestion control

[ducdeptrai1234]

I saw many choices from kamoj congestion control, and the default is westwood. Have anyone try different configs from the page? Is it worth changing the default setting?

 

[L&LD]

Test and report.

 

[microchip]

I use YeAH as CC; Westwood is too aggressive and greedy, though it works well in wireeless networks. All the research papers I've read report roughly the same thing

 

[Droidrat]

Fifty Shades of Congestion Control: A Performance and Interactions Evaluation - 1903.03852.pdf (arxiv.org)

 

[ducdeptrai1234]

I have observed 3 modes: westwood, cubic and yeah.

My internet speed is 220 Mbps, I also used tri-band wireless extender (EX7500). The speed test I use is speedtest.net; the test device is 2015 Macbook Pro 13"; the test location is 20ft apart from the main router, with the wireless extender 10ft from the router. (The location is quite far from house as it is my home workspace). The house has total of 24 devices, with IoT electric plugs, 2 Chromecasts.

The speedtest results are recorded when I connected to the wireless extender, not to the main router. Also, all the test happens in regular evening, which is quite busy period.

+ Device position: R7800 -> 2 wall ->EX7500 -> 2 wall -> laptop.
+ Westwood: 200 Mbps speed, 15 ms latency
+ Yeah, cubic: ~160 Mbps, 12 ms latency

When testing the westwood, some hiccups happens with other devices. With yeah and cubic, no hiccups happens. Also, when using Yeah regularly, the download speed goes up and down, not consistant as westwood.

 

[microchip]

I have observed 3 modes: westwood, cubic and yeah.

My internet speed is 220 Mbps, I also used tri-band wireless extender (EX7500). The speed test I use is speedtest.net; the test device is 2015 Macbook Pro 13"; the test location is 20ft apart from the main router, with the wireless extender 10ft from the router. (The location is quite far from house as it is my home workspace). The house has total of 24 devices, with IoT electric plugs, 2 Chromecasts.

The speedtest results are recorded when I connected to the wireless extender, not to the main router. Also, all the test happens in regular evening, which is quite busy period.

+ Device position: R7800 -> 2 wall ->EX7500 -> 2 wall -> laptop.
+ Westwood: 200 Mbps speed, 15 ms latency
+ Yeah, cubic: ~160 Mbps, 12 ms latency

When testing the westwood, some hiccups happens with other devices. With yeah and cubic, no hiccups happens. Also, when using Yeah regularly, the download speed goes up and down, not consistant as westwood.

Westwood sucks up all bandwidth, it's why you see no fluctuations in speeds. It's an aggressive, greedy algo. YeAH backs off if there's other (minor) network stuff going on

 

[ducdeptrai1234]

Westwood sucks up all bandwidth, it's why you see no fluctuations in speeds. It's an aggressive, greedy algo. YeAH backs off if there's other (minor) network stuff going on

I agree, I will use yeah as I would trade the max speed with the happiness of my household