Release Asuswrt-Merlin 386.12 is now available for AC models

[Amwjujo]

Me again....
Just a quick update.
Today 2 of my AC68U(No Imesh node) lost wifi connections at the same time and they are located at 2 different addresses. I was in one of the place and from the syslog i can only see that the wifi devices are being disconnected and then they recconect back on ... less my wifi cameras(I did fallowed the advice above and set the 2.4Ghs frequency to 20Mhz, but with no success).
Both of the routers are running Unbound, FlexQos and i have the DNS server set to Cleanbrowsing.
Can any of these trigger this situation or it could be something else. The strange part is it happened at the same time for 2 different devices at 2 different locations.

 

[Tomo]

I went back to version 386.12 a long time ago and the wi-fi problems disappeared, i understand that no changes have been made to the wifi module but evidently there is something else that compromises its functioning. I personally believe that the AC models have been supported beyond the normal development processes and i believe it is correct to move to the new AX models as soon as possible.

 

[askan7]

I went back to version 386.12 a long time ago and the wi-fi problems disappeared, i understand that no changes have been made to the wifi module but evidently there is something else that compromises its functioning. I personally believe that the AC models have been supported beyond the normal development processes and i believe it is correct to move to the new AX models as soon as possible.

I got an ax86s b-stock from asus germany to replace my ac86u. There are some improvements but i'd say they are minor.

2.4ghz wifi is the exact same, same RSSI in exact same locations, and same transfer rate at the same RSSI.
5Ghz is a tiny little better RSSI and range. The only advantage i'd say is the 10% throughput ax clients have over AC and 160mhz channels.

UI navigation is faster, but like is it even relevant?
Latencies wired/wireless are the exact same. Ofc you'll have lower latencies at 160mhz vs 80mhz if you're close to maxing out the link, but during normal use latencies are the exact same.
Pinging the router gives me 0.8-1ms wired and ~1.5ms wireless. It seems to be delaying ICMP packets and adding ~1ms on average because ac86u is 0.1ms wired and 0.8ms wireless. Actual latency between devices and to the outside is not affected by this.

 

[Amwjujo]

Yeah, I don't really want to spend money right now upgrading all the routers I have because some cameras that i have are playing funny on the 2.4Ghz. So far from my observations the 5Ghz is doing just fine. It must be something that triggers this behaviour.

 

[Amwjujo]

What is the meaning of this?

Jan 19 14:08:17 kernel: Flags: nZCv IRQs on FIQs on Mode USER_32 ISA ARM Segment user
Jan 19 14:08:17 kernel: Control: 10c53c7d Table: 055c404a DAC: 00000015
Jan 19 14:08:17 kernel: dcd/16913: potentially unexpected fatal signal 6.
Jan 19 14:08:17 kernel: Pid: 16913, comm: dcd
Jan 19 14:08:17 kernel: CPU: 0 Tainted: P (2.6.36.4brcmarm #1)
Jan 19 14:08:17 kernel: PC is at 0x4012d4cc
Jan 19 14:08:17 kernel: LR is at 0xd10c
Jan 19 14:08:17 kernel: pc : [<4012d4cc>] lr : [<0000d10c>] psr: 60000010
Jan 19 14:08:17 kernel: sp : be3ffda8 ip : 000a19c0 fp : be3ffe6c
Jan 19 14:08:17 kernel: r10: 401de020 r9 : 401de3dc r8 : 00000000
Jan 19 14:08:17 kernel: r7 : 0000008e r6 : bedcec28 r5 : 401d9d74 r4 : 00000000
Jan 19 14:08:17 kernel: r3 : 00000000 r2 : 00000000 r1 : be3ffdc8 r0 : fffffdfe

 

[visortgw]

What is the meaning of this?

Jan 19 14:08:17 kernel: Flags: nZCv IRQs on FIQs on Mode USER_32 ISA ARM Segment user
Jan 19 14:08:17 kernel: Control: 10c53c7d Table: 055c404a DAC: 00000015
Jan 19 14:08:17 kernel: dcd/16913: potentially unexpected fatal signal 6.
Jan 19 14:08:17 kernel: Pid: 16913, comm: dcd
Jan 19 14:08:17 kernel: CPU: 0 Tainted: P (2.6.36.4brcmarm #1)
Jan 19 14:08:17 kernel: PC is at 0x4012d4cc
Jan 19 14:08:17 kernel: LR is at 0xd10c
Jan 19 14:08:17 kernel: pc : [<4012d4cc>] lr : [<0000d10c>] psr: 60000010
Jan 19 14:08:17 kernel: sp : be3ffda8 ip : 000a19c0 fp : be3ffe6c
Jan 19 14:08:17 kernel: r10: 401de020 r9 : 401de3dc r8 : 00000000
Jan 19 14:08:17 kernel: r7 : 0000008e r6 : bedcec28 r5 : 401d9d74 r4 : 00000000
Jan 19 14:08:17 kernel: r3 : 00000000 r2 : 00000000 r1 : be3ffdc8 r0 : fffffdfe

Well known "dcd Tainted" kernel crash. Usually caused by Trend Micro (i.e., AiProtection) code.

 

[Amwjujo]

Thank you. I just flashed the firmware again.
Should I just turn off Adaptive qoS? Can this cause the problems with my 2.4Ghz and not being able to connect certain IoT devices to my network .. or getting randomly disconnected?

 

[bibikalka]

I got an ax86s b-stock from asus germany to replace my ac86u. There are some improvements but i'd say they are minor.

...

It seems this router has the very same cpu as AC86u, and same RAM size. It appears to be a simple bump of AC to AX for AC86u!

So even faster interface is a bit suspect - could it be simply due to the newer firmware series?

 

[askan7]

It seems this router has the very same cpu as AC86u, and same RAM size. It appears to be a simple bump of AC to AX for AC86u!

So even faster interface is a bit suspect - could it be simply due to the newer firmware series?

Yes. 388 vs 386.

CPU is irrelevant, 2 vs 4 of the same core makes 0 difference in speed.
Ram only matters if u run a lot of services/scripts. I have no issues with 512mb but I don't run VPN or aiprotection

 

[visortgw]

Yes. 388 vs 386.

CPU is irrelevant, 2 vs 4 of the same core makes 0 difference in speed.
Ram only matters if u run a lot of services/scripts. I have no issues with 512mb but I don't run VPN or aiprotection

Incorrect! Multiple cores can process multiple threads in parallel. Also, RT-AX68U has one 2.5.Gbps LAN/WAN port, which can make huge difference in Internet throughput EVEN IF YOUR ISP SPEED IS LESS THAN 1 GBPS. AX routers, in general, provide much higher WiFi bandwidth. GT-AX6000 or RT-AX86U Pro improve overall performance even more.

 

[askan7]

Incorrect! Multiple cores can process multiple threads in parallel. Also, RT-AX68U has one 2.5.Gbps LAN/WAN port, which can make huge difference in Internet throughput EVEN IF YOUR ISP SPEED IS LESS THAN 1 GBPS. AX routers, in general, provide much higher WiFi bandwidth. GT-AX6000 or RT-AX86U Pro improve overall performance even more.

Most things are single threaded. U can see cpu usage in gui. NAT is hardware accelerated unless u run vpn. LAN transfers are hw accelerated even with vpn.
Soo unless you're like using usb sharing and vpn at same time i don't see how 4 vs 2 cores would make a difference.

AX vs AC is like +10-20% mostly due to 1024qam and a bit better hardware/radios. I'd say it's only 10% better at ~10m where u can't use 1024qam. Closer than 10m it's ~20% better due to 1024qam.
But ofc if you comparing 2x2 80mhz ac vs 2x2 160mhz AX, ax will be much faster. The thing is there is a lot of 160mhz ac routers/clients aswell

 

[bibikalka]

Most things are single threaded. U can see cpu usage in gui. NAT is hardware accelerated unless u run vpn. LAN transfers are hw accelerated even with vpn.
Soo unless you're like using usb sharing and vpn at same time i don't see how 4 vs 2 cores would make a difference.

AX vs AC is like +10-20% mostly due to 1024qam and a bit better hardware/radios. I'd say it's only 10% better at ~10m where u can't use 1024qam. Closer than 10m it's ~20% better due to 1024qam.
But ofc if you comparing 2x2 80mhz ac vs 2x2 160mhz AX, ax will be much faster. The thing is there is a lot of 160mhz ac routers/clients aswell

It's a bit strange. You take a newer router with the same CPU clock, and run small tests to see if it's any better than the older router with the same CPU clock. And it turns out it's just a tiny bit better due to the next gen WiFi, and perhaps some interface improvements. Then you argue that the newer radio should not even matter much.

What was exactly that you expected???

I can understand if you saturated both routers with tons of AC or AX clients, and measured the total throughput. But you did not do such extensive testing.

For selfish reasons, I would like @RMerlin to swap radio drivers from AC86u into 388 firmware, and release that - since other than the radio hardware appears pretty much identical. This way one could have AC86U as a sort of AX86s "lite". But, chances of this happening are not very significant !

 

[askan7]

It's a bit strange. You take a newer router with the same CPU clock, and run small tests to see if it's any better than the older router with the same CPU clock. And it turns out it's just a tiny bit better due to the next gen WiFi, and perhaps some interface improvements. Then you argue that the newer radio should not even matter much.

What was exactly that you expected???

I can understand if you saturated both routers with tons of AC or AX clients, and measured the total throughput. But you did not do such extensive testing.

For selfish reasons, I would like @RMerlin to swap radio drivers from AC86u into 388 firmware, and release that - since other than the radio hardware appears pretty much identical. This way one could have AC86U as a sort of AX86s "lite". But, chances of this happening are not very significant !

You misunderstood me.
What I mean is ac vs ax is not a big deal. But the better hardware(radio) combined with 160mhz channels make significantly better.
If it was 80mhz only, ac vs ax it would have a been a really small diference. Most of the gains from AX come from 1024qam and it requires u to be fairly close to the router and low interference.

 

[heysoundude]

Me again....
Just a quick update.
Today 2 of my AC68U(No Imesh node) lost wifi connections at the same time and they are located at 2 different addresses. I was in one of the place and from the syslog i can only see that the wifi devices are being disconnected and then they recconect back on ... less my wifi cameras(I did fallowed the advice above and set the 2.4Ghs frequency to 20Mhz, but with no success).
Both of the routers are running Unbound, FlexQos and i have the DNS server set to Cleanbrowsing.
Can any of these trigger this situation or it could be something else. The strange part is it happened at the same time for 2 different devices at 2 different locations.

Is it possible that unbound (your caching, recursive DNS server) and cleanbrowsing arent cooperating? I'd start there

 

[Amwjujo]

Is it possible that unbound (your caching, recursive DNS server) and cleanbrowsing arent cooperating? I'd start there

You are 100% right. I did turned off cleanbrowsing and it seems to work now ... I'll keep an eye over the weekend.
From what've read here they were supposed to work together.
The strange part is that the 5Ghz network is ok ... only the 2.4Ghz is loosing connection or says "No internet". I also reverted to 386.11 and now I am trying to understand if it was a DNS issue or is something wrong with the 2.4Ghz in the last version or both
Thank you

 

[askan7]

This makes sense.

I wonder how many use cases out there where the distances are short, so the highest rate WiFi can be used. Marketing needs some bullet points, but I laugh when people talk about buffering issues on 1+Gbps connections, and consider upgrading the connection speed further!

This is nice, i didn't know u could change region. For me, in Europe (routers have Germany region). I'm limited to 200mw in the lower 5ghz channels. 1000mw on 100ch+ but I feel performance is actually worse, improves only at the end of the range and at that range it's better to just use a node/ap.

 

[visortgw]

This is nice, i didn't know u could change region. For me, in Europe (routers have Germany region). I'm limited to 200mw in the lower 5ghz channels. 1000mw on 100ch+ but I feel performance is actually worse, improves only at the end of the range and at that range it's better to just use a node/ap.

As long as you don't get caught, that is!

 

[askan7]

As long as you don't get caught, that is!

If I lived in the city center I wouldn't do it for sure. But I don't soo I don't care. I'll try it, since the allowed 1000mw in ch100+ gives me lower performance i'm curious how 1000mw would feel in ch36.

Also no need to for DFS at 80mhz, that's a huge plus.

 

[thiggins]

@bibikalka Information on illegal Tx power adjustment has been deleted...again. Post it one more time and you'll be banned.

 

[bibikalka]

@bibikalka Information on illegal Tx power adjustment has been deleted...again. Post it one more time and you'll be banned.

Oh, sorry about that, did not know ... Sometimes routers are used in places where they were not bought - I got a used one and it seems to say Canada vs US.