[Release] FreshJR Adaptive QOS (Improvements / Custom Rules / and Inner workings)

[Therion87]

This may be a stupid question, but is there a delay in this script becoming active? I am trying to dial in my QOS Bandwidth limits and it seems like when I change a setting, QOS becomes disabled completely.

I was set at 240 and 20 and was tweaking the UL bandwidth to 22 and now DSL Reports Speedtest shows that my connection is at full bandwidth 300/25. I also saw this behavior when I first enabled the script.

5 minutes.

 

[L&LD]

This may be a stupid question, but is there a delay in this script becoming active? I am trying to dial in my QOS Bandwidth limits and it seems like when I change a setting, QOS becomes disabled completely.

I was set at 240 and 20 and was tweaking the UL bandwidth to 22 and now DSL Reports Speedtest shows that my connection is at full bandwidth 300/25. I also saw this behavior when I first enabled the script.

5 minutes is the bare minimum and works for most. I leave it for 10 minutes after any changes to QoS settings or the script itself or at least until the CPU cores have settled back down for a minute or more (when the rest of the network is idle, of course).

 

[Fnz]

5 minutes is the bare minimum and works for most. I leave it for 10 minutes after any changes to QoS settings or the script itself or at least until the CPU cores have settled back down for a minute or more (when the rest of the network is idle, of course).

Thank you for confirming.

 

[Grimluhn]

Disabling built-in rules is not available yet. (It may be implemented in a future release).

Do you anticipate the few hardcoded rules causing poor QoS performance?

(I can provide a guide how to manually remove them)

If you could create a guide that would be amazing for us hardcore gamyers. I'm not that tech savvy but I'm pretty sure I could handle a guide if you make one. Would Appreciate it

 

[Sinner]

Thank you for confirming.

There is a 5min delay in the freshjr qos after enabling or applying changes thats built into the script. Any changes take 5min to actually take effect.

 

[wizin]

Did you try this script with it’s default recommended settings?

It is significantly different from Asus’, botched, implementation of adaptiveQOS.

I tried yours too, see screenshots - the high numbers are when I am not using QOS at all, which made me think should I use this or I am using the wrong settings. I tried sq and fq, played with those bandwidth numbers too and it didnt make any difference

 

[Sinner]

I tried yours too, see screenshots - the high numbers are when I am not using QOS at all, which made me think should I use this or I am using the wrong settings. I tried sq and fq, played with those bandwidth numbers too and it didnt make any difference

not sure which router you have but on my AC3100 250-300mbs is about the highest thruput your going to see I think with qos enabled. Might be a lil higher with 86u or the AX series. Its very cpu intensive along with vpn. Try running a speed test while watching cpu im pretty sure they'll be maxing out.

 

[L&LD]

not sure which router you have but on my AC3100 250-300mbs is about the highest thruput your going to see I think with qos enabled. Might be a lil higher with 86u or the AX series. Its very cpu intensive along with vpn. Try running a speed test while watching cpu im pretty sure they'll be maxing out.

On my RT-AC3100 with QoS enabled, I could get in the 750 Mbps or slightly better range up/down speeds on a 1Gbps connection.

 

[Vexira]

I don't know if it's just me but I seems that chasing QoS priority in the Asus app causes it to reset to automatic bandwidth

 

[Sinner]

On my RT-AC3100 with QoS enabled, I could get in the 750 Mbps or slightly better range up/down speeds on a 1Gbps connection.

You could be right actually. I may have been thinking of 68u stats. It was early and untested for me as my internet isnt that high.

 

[heysoundude]

I tried yours too, see screenshots - the high numbers are when I am not using QOS at all, which made me think should I use this or I am using the wrong settings. I tried sq and fq, played with those bandwidth numbers too and it didnt make any difference

May I make a few suggestions?
QoS makes a rather striking difference to your network's zippity-doodah. As I've learned from my flight lessons, you often forget steps required in a procedure unless you follow the checklist religiously, so refer to the checklist religiously and things go as they're supposed to.

First, please go back and re-read post #2 of this thread, and then post #1. Print them out so you don't have to switch between tabs in a browser when you're tweaking, and highlight sections that you might tend to gloss over.

Second, please realize that it's quite unlikely your connection consistently achieves the speeds you pay your ISP for. Your jpg1 indicates that you have one of the speeds set at 1000Mbps, and your jpg3 shows a high of 850-something (IIRC, which is probably a realistic, if not average, speed); re-reading post #2 should show you that your settings should take into account these actual results. (IE- set your up/down limits at 770-780-790/850-860, save and restart -that means turning QoS off when you make changes, then turning it back on after adjusting them/saving - the saving part is what people overlook/forget most often - and then maybe even reboot the whole shebang, possibly including your modem, to give everything a little kick in their backsides: If there's some sort of logic problem or conflict from tweaking things, it usually gets sorted with a reboot)

 

[L&LD]

I tried yours too, see screenshots - the high numbers are when I am not using QOS at all, which made me think should I use this or I am using the wrong settings. I tried sq and fq, played with those bandwidth numbers too and it didnt make any difference

You are using sfq, it should be fq_codel in 1-jpg.

@heysoundude has some good advice if you care to follow it.

 

[lilstone87]

You could be right actually. I may have been thinking of 68u stats. It was early and untested for me as my internet isnt that high.

What you are thinking of is when HW acceleration is completely disabled. In that case... The RT-AC3100 has around a 250-270mbps max. With adaptive QoS, it doesn't fully disable HW acceleration, and so 800+mbps is possible still.

 

[Marin]

Cannot get dslreports.com to work like it used to in Safari and Chrome to do some testing when I tweak my upload/download settings. I see this happening in my laptop and PC. It was reported sometime last month that the site was having issues.

Does anyone know if they are still having issues or it simply my browser?

Thanks!

 

[wizin]

May I make a few suggestions?
QoS makes a rather striking difference to your network's zippity-doodah. As I've learned from my flight lessons, you often forget steps required in a procedure unless you follow the checklist religiously, so refer to the checklist religiously and things go as they're supposed to.

First, please go back and re-read post #2 of this thread, and then post #1. Print them out so you don't have to switch between tabs in a browser when you're tweaking, and highlight sections that you might tend to gloss over.

Second, please realize that it's quite unlikely your connection consistently achieves the speeds you pay your ISP for. Your jpg1 indicates that you have one of the speeds set at 1000Mbps, and your jpg3 shows a high of 850-something (IIRC, which is probably a realistic, if not average, speed); re-reading post #2 should show you that your settings should take into account these actual results. (IE- set your up/down limits at 770-780-790/850-860, save and restart -that means turning QoS off when you make changes, then turning it back on after adjusting them/saving - the saving part is what people overlook/forget most often - and then maybe even reboot the whole shebang, possibly including your modem, to give everything a little kick in their backsides: If there's some sort of logic problem or conflict from tweaking things, it usually gets sorted with a reboot)

Thanx

Changed to fq_model and made manual bandwidth to 850/850 results from

https://www.speedtest.net/result/8242545972

https://i.imgur.com/zIH99WZ.png

 

[Fnz]

Thanx

Changed to fq_model and made manual bandwidth to 850/850 results from

https://www.speedtest.net/result/8242545972

https://i.imgur.com/zIH99WZ.png

Your results show 916. Yet you are limiting to 850? Did you wait 5 minutes before testing? Each change disables QOS completely for around 5 minutes.

 

[heysoundude]

Your results show 916. Yet you are limiting to 850? Did you wait 5 minutes before testing? Each change disables QOS completely for around 5 minutes.

Right! it looks like he can creep the number up a smidge to try to come closer to what his ISP sells him, but the pipe is dynamic, after all: the fine print likely reads something to the effect of "maximum theoretical speeds".
(the saving/switching/rebooting I suggested that got them there was supposed to give them that ~5mins...)

 

[Fnz]

Right! it looks like he can creep the number up a smidge to try to come closer to what his ISP sells him, but the pipe is dynamic, after all: the fine print likely reads something to the effect of "maximum theoretical speeds".
(the saving/switching/rebooting I suggested that got them there was supposed to give them that ~5mins...)

What I was getting at is that the results he posted are likely with QOS off. If he set to 850, and did not wait the 5 minutes to test, they would look like that. Wait 5 minutes, then test at DSL Reports for Buffer Bloat and speed.

 

[CriticJay]

Cannot get dslreports.com to work like it used to in Safari and Chrome to do some testing when I tweak my upload/download settings. I see this happening in my laptop and PC. It was reported sometime last month that the site was having issues.

Does anyone know if they are still having issues or it simply my browser?

Thanks!

Yes, they're still having lots of issues showing the "grade" of the speedtest/bufferbloat.

You know what? If you live in North America, using "fast.com" as a sort of bufferbloat test works well.
With QoS disabled, the "loaded" latency for fast.com should be very high - like, over 100ms for most Cable Modem connections.
With QoS enabled (especially a good QoS, like Merlin firmware + Fresh's script), the "loaded" latency should be no more than 15ms worse than your "unloaded" latency.

Try it out.

 

[wizin]

Your results show 916. Yet you are limiting to 850? Did you wait 5 minutes before testing? Each change disables QOS completely for around 5 minutes.

Yes waited, actually after more testing it seems sq model is better for me

thanx