CakeQOS CakeQoS-Merlin v2.1.1

[kuki68ster]

Oops silly me doh !...

My line speed is 80/20 realistically I get out of that 74mps download and 19mps upload when doing a speed test so that's pretty darn good . With cake on I can get around 74.5mps and still 19mps .

Do a test on https://www.waveform.com/tools/bufferbloat, to see if you are getting an A+ with your actual config...

For PPPoE connections, setting the MTU to 1492 bytes is generally recommended. This accounts for the PPPoE overhead. (According to chatgpt)...

If you suspect MTU issues:

Ping Test:

• Use the ping command with the -f (don't fragment) option and different packet sizes to find the largest MTU that doesn’t fragment. For example: ping -f -l 1472 google.com
• Adjust the MTU value based on the results from this test.

Check for Fragmentation:

• Ensure that packets are not being fragmented. Fragmentation can indicate that the MTU setting is too high.

 

[marcd5123]

Do a test on https://www.waveform.com/tools/bufferbloat, to see if you are getting an A+ with your actual config...

For PPPoE connections, setting the MTU to 1492 bytes is generally recommended. This accounts for the PPPoE overhead. (According to chatgpt)...

If you suspect MTU issues:

Ping Test:

• Use the ping command with the -f (don't fragment) option and different packet sizes to find the largest MTU that doesn’t fragment. For example: ping -f -l 1472 google.com
• Adjust the MTU value based on the results from this test.

Check for Fragmentation:

• Ensure that packets are not being fragmented. Fragmentation can indicate that the MTU setting is too high.

Hi Buddy thanks for that much appreciated!

Yes Oddly enough I have my MTU set as 1492 on the Asus RT-AX86U Pro settings it self actually EE UK technical told me they have it set to 1492 and it set it automatically once I put in my configuration when putting in my username as password as I have the 86U Pro set as "router" mode to my previous TP-Link VR9600 router which is set to bridge-mode (dumb ADSL2 modem) so I know that is set 100%. The TP-Router has a very high quality Braodcom chip set so I decided to use that still as my dumb modem why not it's better than the cheapo ISP free router they give out lol .I purchased that back in 2015 and serviced it a few times now .

I then did a Ping test on my laptop to make sure I was not getting any fragmentation what so ever so I did a Ping test to Google 8.8.8.8 and Cloudfare servers and it was absolutely spot on ..
I started at 1400 and worked upwards 1450...1460...ext.ext then ...

1464+28 for the header which gives me 1492 and all packets were returned unfragmented bingo . I ran this again only a few months back to make sure as I had already perform this test last year.

So do I put 1492 were it states MPU ???....

I will also look at the other information you have Kindly provided I presume I leave it at "PTM" as it's ADSL2 FTTC .

Many thanks

 

[kuki68ster]

Hi Buddy thanks for that much appreciated!

Yes Oddly enough I have my MTU set as 1492 on the Asus RT-AX86U Pro settings it self actually EE UK technical told me they have it set to 1492 and it set it automatically once I put in my configuration when putting in my username as password as I have the 86U Pro set as "router" mode to my previous TP-Link VR9600 router which is set to bridge-mode (dumb ADSL2 modem) so I know that is set 100%. The TP-Router has a very high quality Braodcom chip set so I decided to use that still as my dumb modem why not it's better than the cheapo ISP free router they give out lol .I purchased that back in 2015 and serviced it a few times now .

I then did a Ping test on my laptop to make sure I was not getting any fragmentation what so ever so I did a Ping test to Google 8.8.8.8 and Cloudfare servers and it was absolutely spot on ..
I started at 1400 and worked upwards 1450...1460...ext.ext then ...

1464+28 for the header which gives me 1492 and all packets were returned unfragmented bingo . I ran this again only a few months back to make sure as I had already perform this test last year.

So do I put 1492 were it states MPU ???....

I will also look at the other information you have Kindly provided I presume I leave it at "PTM" as it's ADSL2 FTTC .

Many thanks

Check this:

https://www.snbforums.com/threads/wan-packet-overhead.74752/

Cake Built-in Setup, Overhead MPU speed limits

Hi, I'm so glad merlin has cake Built-in in latest alpha build and it's awesome for HND users, it's a one time setup and magic done thingy, Thanks to @RMerlin I've FTTH(fiber to home) GPON Connection and fiber going direct to ISP modem and ISP modem is obviously in bridge mode and PPPOE...

www.snbforums.com

I think you can leave it to zero (default), if you are already getting an A+ on the bufferbloat test...

 

[marcd5123]

Check this:

https://www.snbforums.com/threads/wan-packet-overhead.74752/

Cake Built-in Setup, Overhead MPU speed limits

Hi, I'm so glad merlin has cake Built-in in latest alpha build and it's awesome for HND users, it's a one time setup and magic done thingy, Thanks to @RMerlin I've FTTH(fiber to home) GPON Connection and fiber going direct to ISP modem and ISP modem is obviously in bridge mode and PPPOE...

www.snbforums.com

I think you can leave it to zero (default), if you are already getting an A+ on the bufferbloat test...

Hi I just ran the test a few times and it seems to settle on need results. . Thanks .

 

[kuki68ster]

Hi I just ran the test a few times and it seems to settle on need results. . Thanks .

Did you follow the advice:

Disable any bandwidth-heavy tasks on your network

It’s important that your Internet connection’s upload and download connections aren’t saturated already.

Make sure no devices are currently uploading or downloading large files from the Internet (e.g. backup/sync tasks by Google Drive, Dropbox, etc).

It seems something is not ok with your config...

 

[marcd5123]

Did you follow the advice:



It seems something is not ok with your config...

Well I ran it using my Oppo Find X5 Pro using WiFi 6 AX and it's on channel 116 (UK) theirs absolutely no one else on the channel but me .

Nothing else is on using the network but my phone and it's been rock solid for well over 1 week now . I dont seem to get any buffering issues when streaming and it's virtually instantaneous streaming 4K content.

What should I check that could be causing issues.. I ran speed test also via the router it self and it gives the same download and upload speeds and that was even before Cake Qos was switched on so it's more or less giving the same speeds .

I also just ran this from the Asus app and the speed test from Asus router directly and both give the same results .

I'm based in Uxbridge -West London.

 

[Yakumo]

Well I ran it using my Oppo Find X5 Pro using WiFi 6 AX and it's on channel 116 (UK) theirs absolutely no one else on the channel but me .

Nothing else is on using the network but my phone and it's been rock solid for well over 1 week now . I dont seem to get any buffering issues when streaming and it's virtually instantaneous streaming 4K content.

What should I check that could be causing issues.. I ran speed test also via the router it self and it gives the same download and upload speeds and that was even before Cake Qos was switched on so it's more or less giving the same speeds .

I also just ran this from the Asus app and the speed test from Asus router directly and both give the same results .

I'm based in Uxbridge -West London.

Cake can't really work unless you manually set the ul and dl bandwidth on the QoS page, 5-10% lower than your tested speeds. Do that and then run the bufferbloat test again. That is more important than any of the other settings really. With the bandwidth lowered enough and the line not otherwise being used when running the bufferbloat test (the test itself it will use it enough) then you should start to see only +1 or 2ms, and A+ ratings.

Try 70 down 17 up to start and if not getting results that good lower it a little more, it is sacrificing peak speeds for better user sharing.

 

[marcd5123]

Cake can't really work unless you manually set the ul and dl bandwidth on the QoS page, 5-10% lower than your tested speeds. Do that and then run the bufferbloat test again. That is more important than any of the other settings really. With the bandwidth lowered enough and the line not otherwise being used when running the bufferbloat test (the test itself it will use it enough) then you should start to see only +1 or 2ms, and A+ ratings.

Try 70 down 17 up to start and if not getting results that good lower it a little more, it is sacrificing peak speeds for better user sharing.

Hiya ,Thank you for that much appreciated ... However slightly confused I understand what you are saying but if that is the case then why is their a "automatic" setting surly that should be enough and let the Program-addon run it self...??

Then why have that in place it would make no scene ...

 

[kuki68ster]

Cake can't really work unless you manually set the ul and dl bandwidth on the QoS page, 5-10% lower than your tested speeds. Do that and then run the bufferbloat test again. That is more important than any of the other settings really. With the bandwidth lowered enough and the line not otherwise being used when running the bufferbloat test (the test itself it will use it enough) then you should start to see only +1 or 2ms, and A+ ratings.

Try 70 down 17 up to start and if not getting results that good lower it a little more, it is sacrificing peak speeds for better user sharing.

@marcd5123 do this...

Start from scratch so you are sure that everything is installed as default...

Before installing anything do a few tests (speed/bufferbloat) just to check your default speeds (without cake activated) and then activate cake, configure as suggested with 5% to 10% of your lowest result of previous speed tests and do a bufferbloat test again to check if the result is what you expect...

 

[kuki68ster]

Hiya ,Thank you for that much appreciated ... However slightly confused I understand what you are saying but if that is the case then why is their a "automatic" setting surly that should be enough and let the Program-addon run it self...??

Then why have that in place it would make no scene ...

You will get a better result if configured manually...

Just do a few speeds tests and input the lowest result...

 

[marcd5123]

@marcd5123 do this...

Start from scratch so you are sure that everything is installed as default...

Before installing anything do a few tests (speed/bufferbloat) just to check your default speeds (without cake activated) and then activate cake, configure as suggested with 5% to 10% of your lowest result of previous speed tests and do a bufferbloat test again to check if the result is what you expect...

As stated above then why have a "automatic" setting?? I

 

[kuki68ster]

As stated above then why have a "automatic" setting?? I

Give it a read:

CakeQOS Settings for AC86U?

Hello guys. So I was interested in using CakeQOS for my bad connection (50D/5U) with 2-3% packet loss and high bufferbloat. So I wanted to know if somebody can point me to any useful guides on how I could setup CakeQOS on my Ac86u and find the best settings possible for my specific occasion.

www.snbforums.com

The basic, is that has @Tech9 said:

Tests and experience. Automatic means no bandwidth limits. Adjusting upload/download limits manually and below 100% CPU load works best. On AC86U that means below 300Mbps. For faster ISP lines Cake (the CPU) will limit WAN-LAN throughput. Cake is incompatible with Runner/Flow Cache.

 

[dave14305]

As stated above then why have a "automatic" setting?? I

Automatic setting was originally for the other QoS types. When cake was added, it came to mean “bandwidth unlimited” so cake will not try to control the bandwidth, but only the fair consumption of unknown bandwidth. Don’t use Automatic.

 

[marcd5123]

@marcd5123 do this...

Start from scratch so you are sure that everything is installed as default...

Before installing anything do a few tests (speed/bufferbloat) just to check your default speeds (without cake activated) and then activate cake, configure as suggested with 5% to 10% of your lowest result of previous speed tests and do a bufferbloat test again to check if the result is what you expect...

Theirs nothing installed...I switched it on and had surprising benefits to my streaming experience I've stated this all ready with my previous posts so don't want to repeat myself sorry buddy .

Then why have a "automatic" setting the entire idea is let the addon-program handle everything and I feel it is ..just because some website States their "bufferfloat" yet everything else's works like a dream I'm very sceptical about that website.

Running multiple speed tests directly on the router itself and using the Asus app on my mobile States otherwise with regards to latency.

Yes I "may" look into it ...and perhaps run it manually I'm going to do some more reading/research first before starting to sacrifice some of my bandwidth yet their is a "automatic" setting which is engaged....thank you buddy .

 

[marcd5123]

Automatic setting was originally for the other QoS types. When cake was added, it came to mean “bandwidth unlimited” so cake will not try to control the bandwidth, but only the fair consumption of unknown bandwidth. Don’t use Automatic.

Hiya okay I'm going to look and read some more into this with the links provided thank you and try some of the settings stated by other members and your self .

i e 70mbps down and 17mbps up m see how that runs see how this goes ...this should prove interesting

I presume I leave "MPU" at default "0" ???

 

[marcd5123]

Cake can't really work unless you manually set the ul and dl bandwidth on the QoS page, 5-10% lower than your tested speeds. Do that and then run the bufferbloat test again. That is more important than any of the other settings really. With the bandwidth lowered enough and the line not otherwise being used when running the bufferbloat test (the test itself it will use it enough) then you should start to see only +1 or 2ms, and A+ ratings.

Try 70 down 17 up to start and if not getting results that good lower it a little more, it is sacrificing peak speeds for better user sharing.

I presume I leave "MPU" at 0 default??? I won't touch this.

 

[marcd5123]

Hiya everyone Okay update from me....Okay okay you guys were correct I hold my hands up ..

It seriously worked with what you all said .

So this is based using my mobile phone which is Oppo Find X5 Pro which is AX compatible WiFi 6.

Using DFS channal 116 @ 80mhz (Asus RT-AX86U Pro) latest MerlinWRT software well the 2nd lastest not the VPN Kill switch update .

I ran the test 3 times to check for absolute consistency and each of this 3 times came back A+ that's a huge difference from before!!!!!.

I based it on using 10% bandwidth so 70mps DL and 17mps UL here's the screenshots I took .

Let's see what happens tonight when I use these settings using my Nvidia Shield Pro 2019 which has a Kodi Diggz Build and RD account.

Also my Apple TV 128GB device.

So my understanding with this now in place "correctly" 100% setup it should seriously optimise my bandwidth and really smooth out any kinks as such .I will monitor this over the coming week.

Thanks once again for all your help even if I was rather sceptical .

 

[marcd5123]

Hi guys I forgot to wright in the above...

Do I leave MPU at "0" default and should I change it from PTM to Normal???

 

[kuki68ster]

Hi guys I forgot to wright in the above...

Do I leave MPU at "0" default and should I change it from PTM to Normal???

Leave it with the config that you

Hi guys I forgot to wright in the above...

Do I leave MPU at "0" default and should I change it from PTM to Normal???

if you already are getting an A+ leave it...

 

[marcd5123]

Leave it with the config that you

if you already are getting an A+ leave it...

Hi Thanks for that ...I'm going to run further tests. Now I know it's working correctly with MPU at 0 (default) and Mode at "PTM" which is for VDSL2 ...

I'm going to see what results I get with changing Mode from PTM to Normal if better results then I'll keep it if the same then I'll change it back to PTM .

I'm also going to tweek very slowly the DL rate as it's set for 10% I'm going to try and test with just 5% and see if I get same results or worse results.

I'm also going to do the obvious and switch CakeQos off and see what the bufferfloat test comes up with as well the previous "bad" results was with CakeQos set to "automatic".

I get real world 74mps consistent connection out of my 80mps which is pretty darn good as that's a FTTC ..and 18.5/19 from a 20UL . Without any Qos in place .

So if I can tweek carefully possibly around 5-7% perhaps I can get close to bang on 70mps with CakeQos on .

I'll post my results when done Wish me luck .



I'll post my results.