No, I have not.
Searching I found this message.
Is this still valid?
Should I also add the nlbwmon fix?
For the WiFi I will test again tomorrow, but I think that it didn’t work even with 80 MHz on channel 36. If this is the case what should I do? Should I flash again with a sysupgrade or via serial?
Thanks in advance
Yes
No
Do a sysupgrade, serial do the same for this.
What was the previous firmware?
I flash twice when I change from nss to non-nss versions and vice versa.
It was latest stock, then I tried installing openwrt today, but I maybe bricked the device since I runned the script flash_to_openwrt not from /tmp/ApplicationData/ and it would not boot (led blinking red).
Then I reverted to stock via serial and I reused the script to install openwrt (this time correctly).
Flashed again and it works.
Thanks
For those using this device and complaining about it being too hot. If you have it running on your desk, the natural air flow is poor. Try mounting it on a wall (preferably with the back side up). It will run much cooler, with less fan noise.
Hey @asvio , quick question if you don't mind.
I know that you have a few branches active, mainly based on main and 24.10.
With main I am using apk-manager in my config. With your 24.10 branch, should I edit the default to opkg manager for it to work correctly (or according to openwrt policy)?
cheers
Of course you can.
But look inside at the last commit i publish on 24.10-rc5 and you can see i changed to make opkg download from openwrt server.
Is it this for what you are looking for?
Hey,
Thanks a lot for clarification!
I built your 24.10 branch (not rc5) this morning and added opkg manually to my config just in case. I wasn't sure if it's correct as config seed was still on apk-manager (they're the same for main and 24.10).
EDIT: I usually glance through commits, but must have missed this one!
cheers
@asvio Instead of running the fan at full speed, is there a way to increase the fan speed gradually as the temp goes up, in order to reduce noise? Does the hardware support that?
PS: @linchpin81 sorry, I didn't mean to reply you. I just clicked on the wrong "reply" button
According the internal photos at FCC website, this device uses FD052510LL-N 5V 25x25x10mm mini axial fan
Fan description at the manufacturer website
Fan Description
The 2510 cooling fan has a built-in sensitive temperature sensor,
which can realize the change of speed and current with temperature.
Today device's fan was turned on without any load on the device. I checked the log and saw these warning messages flooding.
Thu Jan 23 23:50:39 2025 kern.warn kernel: [779403.708317] ath11k_warn: 22 callbacks suppressed
Thu Jan 23 23:50:39 2025 kern.warn kernel: [779403.708365] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:39 2025 kern.warn kernel: [779403.810727] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:39 2025 kern.warn kernel: [779403.913122] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:39 2025 kern.warn kernel: [779404.015522] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:39 2025 kern.warn kernel: [779404.117924] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:39 2025 kern.warn kernel: [779404.322728] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:39 2025 kern.warn kernel: [779404.527526] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:40 2025 kern.warn kernel: [779404.732329] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:40 2025 kern.warn kernel: [779404.937136] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:40 2025 kern.warn kernel: [779405.044627] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:44 2025 kern.warn kernel: [779408.725967] ath11k_warn: 24 callbacks suppressed
Thu Jan 23 23:50:44 2025 kern.warn kernel: [779408.726015] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:44 2025 kern.warn kernel: [779408.828375] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:44 2025 kern.warn kernel: [779408.930772] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:44 2025 kern.warn kernel: [779409.237975] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:44 2025 kern.warn kernel: [779409.340383] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:44 2025 kern.warn kernel: [779409.442778] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:45 2025 kern.warn kernel: [779409.647577] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:45 2025 kern.warn kernel: [779409.749977] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:45 2025 kern.warn kernel: [779409.852382] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:45 2025 kern.warn kernel: [779409.954792] ath11k c000000.wifi: invalid pdev_id 2 in mgmt_rx_event
Thu Jan 23 23:50:51 2025 kern.warn kernel: [779416.201376] ath11k_warn: 26 callbacks suppressed
I did a wifi restart and then service network restart but it did not stop. After restarting the device, it stopped.
I built and ran the nbg7815-main-nss branch of @asvio since 4.9.2024 and it wifi speed was always really fast. On 8.1.2025 I thought lets try the new release candidate branch 24.10.0-rc5-nss. Everything was ok except the wifi down speed. Upload was great, nearly 100 MB/s but down it was between 20 to 40 MB/s. It stayed the same with rc6. So I switched back to main-nss branch today and speed is back to normal. No idea what caused that. Just to let you know.
It's true,
there is a performance problem between wlan and router since May 2024.
Therefore, it affects the services offered by the router: file sharing, iperf3...
The problem does not exist between lan and wlan.
Until now, no one had commented anything....
You are right about that. iperf3 through the router (over Wifi) brings nearly 100 MB/s in both directions. But samba4 or iperf3 hosted on the router brings only 30 to 40 MB/s in down direction. CPU is not the bottle neck (also over wifi). But when you transfer from samba4 through another router it is much higher (via lan 90 MB/s and via ac wifi 60 MB/s). So it looks like this is somehow related to the wifi of the nbg7815.
Are you using 80mhz channel width for 5ghz band?
I get >1800mb/s => 225 MB/s
Well, it depend how many simultaneous connection do you use.
If you don´t use -P parameter with iperf3 you will get something like this:
-P 16 you will get this:This is why I think it seems more like a problem with how the router handles incoming Wi-Fi connections when it has to process them than a problem with the Wi-Fi transmission itself. More connections you have more global download you will have.
This behavior is only present on nss firmware.
Oficial openwrt patched to support 160mhz do not have this "bug" but there is a problem on wlan -> lan connections with a max throughput uploading around 1000mb/s and lots of packets lost during test.
(this is a test wlan-lan made last november, but the problem is still there.)
For reference, this is what you can get from wlan-router connections with non-nss firmware with 160mhz support. (No packet lost and a very good upload speed)
Yes 80 MHz and I measured always using one stream. When I transfer via samba4 I also have always one stream. My goal was to have the same with iperf3 in order to figure out if the problem is related to samba4.
When moving from kernel 6.6.32 to 6.6.47 and now to 6.6.73 I realized that drop in samba4 downspeed. I am not sure if it already was that way with 6.6.47 but with 6.6.32 is was defintely higher. Before I used a r7800 and the higher samba speed was one of the reasons I switched to the nbg7815. On the r7800 it was around 40 MB/s (megabyte per second) from a lukscrypted SSD.
Thanks for sharing these measurements. Your screenshots are impressive. I am looking into this tool.
I'd heard that samba uses more than 1 connection for Windows. This is why you get more download on samba than iperf3.
That's a good news.
What features (except the fan) does the official OpenWRT not have support for NBG7815 at the moment?
What about the NSS?
Don't be mistaken.
I don't think OpenWrt will support 160mhz for this device.
It will all depend on Zyxel releasing a new firmware that includes board2.bin to support this feature and I'm 100% sure they won't. OpenWrt will always use the firmware they include in their firmware.
Regarding NSS it has already been said countless times that there will be no official support.
The features not implemented in the official version besides those mentioned are the LED and fan functionality, everything else is already implemented.. I can't think of anything else that could be missing.
What worries me is that when OpenWRT moved to DSA, R7800 maintainers stopped working on the NSS. I hope Armor G5 does not have a similar fate, I really like this device
Sadly DSA drivers for ipq807x aren't really in sight, yet. But be aware that NSS can't be supported/ maintained by the OpenWrt developers (it's just a massively invasive patch, disregarding mainline kernel standards and circumventing large parts of netfilter), so breaking that is not a consideration for potential future development - and given the size of it, it can break very easily.
If you want to rely on continued performance, it's better to stick to mainline supported (kernel- and OpenWrt) devices and features, so in this case a device that meets your performance expectations without relying on NSS (be it filogic or x86_64 or something else).
Disclaimer: I'm not an OpenWrt developer and obviously can't speak for the project.
Am I understanding correctly that the NSS fork also includes the other Armor G5 specific features, like the fan and the LED control? In that case, I'll probably try it.
By the way, does the fan just never run at all if you use the official build? I've never heard it at least. In that case, I really need to prevent the thing from cooking itself, lol.
