QOS on 1gbps network speeds

[vit5421]

given the experience of the AC86U, it's probably safe to say that all HND-based Asus routers should be able to handle Gigabit speeds with Adaptive QoS enabled

I not able to reach more than 850mbs with Adaptive QoS enable on much more powerful ax88u

 

[nickie]

my ac86u can't reach more than 400/430 dl speed on a 500/100 (usually 510+/105) hfc connection, with adaptative QoS set at a maximum of 500/100. @CriticJay what are your settings?

 

[Tech9]

I not able to reach more than 850mbs with Adaptive QoS enable on much more powerful ax88u

AC86U and AX88U have about the same routing performance. The same Gen CPU with the same cores. One core is used for routing. AX88U is more powerful only if you can use the 4 cores. Few OpenVPN clients running on different cores, as an example.

 

[nickie]

AC86U and AX88U have about the same routing performance. The same Gen CPU with the same cores. One core is used for routing. AX88U is more powerful only if you can use the 4 cores. Few OpenVPN clients running on different cores, as an example.

That's why I don't understand some of the figures being reported... How can some users get +800mbps on ax88u which should have a similar performance as the ac86u? If I recall correctly, even merlin has already stated that, based on the stock firmware, we shouldn't expect more than 400mbps with these routers using adaptive QoS (which is what I'm seeing on my ac86u). I get that FlexQoS seems to have better performance (different/more efficient algorithms?), but the difference in performance is baffling....

 

[L&LD]

The hardware is not the only thing affecting the performance of the routers. The Broadcom SDK's are different too.

This is one reason why the two-year newer RT-AX86U is superior, in actual use, to the on-paper 'superior' RT-AX88U with 'more' hardware inside.

 

[CriticJay]

my ac86u can't reach more than 400/430 dl speed on a 500/100 (usually 510+/105) hfc connection, with adaptative QoS set at a maximum of 500/100. @CriticJay what are your settings?

With Adaptive QoS set to 940 down and 28.3 up, I'm able to achieve sustained speeds of 920 down and 27.3 up.

FlexQoS installed, set to fq_codel. AiProtect disabled. Other configurations listed in signature.

I always use a wired Gigabit-LAN connected PC for speed tests. Eliminate the variability and overhead (WPA2 encryption/decryption etc) of wireless.

 

[CriticJay]

If I recall correctly, even merlin has already stated that, based on the stock firmware, we shouldn't expect more than 400mbps with these routers using adaptive QoS (which is what I'm seeing on my ac86u).

I am not sure he's ever said that.

And FlexQOS does not replace Adaptive QoS - it tweaks it using a variety of commands. You can see which commands by reviewing the script.

Adaptive QoS, like AiProtect, is a black box so presumably only Asus can adjust the source code. I think Merlin had to do some fancy "code injection" stuff to get fq_codel in 384 firmwares.

 

[RMerlin]

That's why I don't understand some of the figures being reported... How can some users get +800mbps on ax88u which should have a similar performance as the ac86u? If I recall correctly, even merlin has already stated that, based on the stock firmware, we shouldn't expect more than 400mbps with these routers using adaptive QoS (which is what I'm seeing on my ac86u). I get that FlexQoS seems to have better performance (different/more efficient algorithms?), but the difference in performance is baffling....

400 Mbps would be with Traditional QoS. These routers should be perfectly capable of hitting 600-800 Mbps with Adaptive QoS.

 

[nickie]

400 Mbps would be with Traditional QoS. These routers should be perfectly capable of hitting 600-800 Mbps with Adaptive QoS.

Thenk you merlin. Using wifi ( I don’t use wired connections), aiprotection and traffic analyser I only get about 400mb. Disabling aiprotection and traffic analyser I get about 450, but that defeats the purpose for which I bought the Ac86u…

 

[Paliv]

Using wifi

I think this answers why you have a disparity.

 

[John DeLuca]

I ended up getting my self a AX11000 and Cm2000 to take advantage of my ISP 1200Mbps plan. I have those in hand now with merlin on the router and will do some testing, if I understood right the best way to test for bufferbloat is to run a speed test and then ping to the host computer using another device correct? should the computer pinging out to the host be connected on via lan for the most accurate results? I also ordered a dual lan card and ill do some testing to see if that changes anything with bufferbloat but I highly doubt it. I will probably do testing with flexqos as I want to still have hardware acceleration.

 

[nickie]

I think this answers why you have a disparity.

Probably.... Although it is a high toll...

 

[Paliv]

Probably.... Although it is a high toll...

What speed do you get on WiFi with no QoS, the full 500/100?

 

[nickie]

What speed do you get on WiFi with no QoS, the full 500/100?

Yep. 515/105, normally.

 

[John Davis]

400 Mbps would be with Traditional QoS. These routers should be perfectly capable of hitting 600-800 Mbps with Adaptive QoS.

ac86u here on 1gbps/0.5gbps fibre connection - with adaptive qos on and flexqos installed I'm hitting 940/470 speedtesting against a local test target, and that's with aiprotection on and a pppoe connection

If I turn adaptive QOS off my download speeds remain the same but upload speed actually DROPS to about 300

 

[nickie]

ac86u here on 1gbps/0.5gbps fibre connection - with adaptive qos on and flexqos installed I'm hitting 940/470 speedtesting against a local test target, and that's with aiprotection on and a pppoe connection

If I turn adaptive QOS off my download speeds remain the same but upload speed actually DROPS to about 300

That’s using Ethernet, right? Have you tried Wi-Fi?

 

[John Davis]

That’s using Ethernet, right? Have you tried Wi-Fi?

fastest wifi device I have is still only 2x2 - speedtest off that is 550/450, but it's almost certain the limit is the device not QOS

 

[avtella]

QoS on gig makes sense if it’s aysmetrical. Like 1.4 Gbps down and 40-50 Mbps Up like I get on Comcast. I use it because at times I do saturate the upload side.

 

[Paliv]

QoS on gig makes sense if it’s aysmetrical. Like 1.4 Gbps down and 40-50 Mbps Up like I get on Comcast. I use it because at times I do saturate the upload side.

I am in the same boat. We never saturate even the QoS limited download, but when working multiple people in our house can saturate the upload. It’s only disappointing when I run a speed test.

 

[Tech9]

I ended up getting my self a AX11000 and Cm2000 to take advantage of my ISP 1200Mbps plan. I have those in hand now with merlin on the router and will do some testing, if I understood right the best way to test for bufferbloat is to run a speed test and then ping to the host computer using another device correct? should the computer pinging out to the host be connected on via lan for the most accurate results? I also ordered a dual lan card and ill do some testing to see if that changes anything with bufferbloat but I highly doubt it. I will probably do testing with flexqos as I want to still have hardware acceleration.

Oh, boy! You don't read carefully or understand the advice folks around are giving you (in both threads). In my eyes you overpaid for a wrong 3-radio (not 3-band) router, your advantages expectations are unclear, you still don't know when bufferbloat happens (if it happens) and on top you purchased extra hardware you don't need. What's the point asking for advice and completely ignoring it?