I haven’t tried that before. I was following the NSS thread all last year and jumped on board after the big stability improvement with fixing the NSS frequency at max frequency.
My three r7800’s are stable (no reboot issues) with performance governor + max NSS freqency.
Mhhh I would ask some test with performance and nss scaling just to confirm a theory. Don't know if you can test this.
I do add these stuf on my ea8500 .and it's working fine (over 15 days ,no reboot ,no freeze)
did you compile the image yourself?
yes,what can I help you?
patch 999-00-Add.....
you need to change
+ qcom,low-frequency = <733000000>; /* orig value 110000000 */
+ qcom,mid-frequency = <733000000>; /* orig value 550000000 */
+ qcom,max-frequency = <733000000>;
to
+ qcom,low-frequency = <110000000>;
+ qcom,mid-frequency = <550000000>;
+ qcom,max-frequency = <733000000>;
Tell me if any reboot occurs.
Remember you need this
echo performance > /sys/devices/system/cpu/cpufreq/policy0/scaling_governor; echo performance > /sys/devices/system/cpu/cpufreq/policy1/scaling_governor
This is to test if the reboot are caused by the cpu scaling or the nss scaling. (since it looks like we always tested cpu + nss scaling but never nss scaling and cpu set to max)
ok,got it
I will let you know asap
cheers
- sdcc1bam:dma@12402000 {
+ nss0: nss@40000000 {
+ compatible = "qcom,nss";
+ qcom,low-frequency = <733000000>; /* orig value 110000000 */
+ qcom,mid-frequency = <733000000>; /* orig value 550000000 */
+ qcom,max-frequency = <733000000>;
......
+ nss1: nss@40800000 {
+ compatible = "qcom,nss";
+ qcom,low-frequency = <733000000>; /* orig value 110000000 */
+ qcom,mid-frequency = <733000000>; /* orig value 550000000 */
+ qcom,max-frequency = <733000000>;
Shoud I change both place? this is from @ACwifidude 's lastest master repo
yes both place
any news ?
Hi,I tested only my ea7500 yesterday with your new patch,it auto-rebooted after 1-2 hours.
I'm testing my ea7500 with @KONG's new patch (without your new patch ).it's working for 1d 2h 47m 15s now.
Should I test ea8500 or r7800 with your new patch later?
thanks
I have used firmware with nss and the perforance governor. and the result is the same. it is the load on the router that causes it to drop, I have no doubts. If I stop using torrent, it lasts longer, and thus, I imagine that by removing services the improvement will increase.
I also tried 600mhz from the companion and same result.
I don't know if the Kong patch would improve this. when you have it on r7800 I test it, although I prefer lighter firmware 
The Kong patch is a tweak to the .dts file for the ea8500.
I’ll rebase this weekend with the ea8500 updates. 
I think it is not applied yet, in the kernel log it appears 800 mhz three times
I've successfully flashed the openwrt-21.02-nss-qsdk10.0 onto my R7800 router and happy with the stability - so thank you @ACwifidude and the others involved in the process
There were few small issues I encountered while building the code:
Only the Asrock G10 image was built, instead of the R7800 I needed. I had to add these two lines to the diffconfig file:
CONFIG_TARGET_PROFILE="DEVICE_netgear_r7800" CONFIG_TARGET_ipq806x_generic_DEVICE_netgear_r7800=y
instead of:
CONFIG_TARGET_DEVICE_ipq806x_generic_DEVICE_netgear_r7800=y
after manually selecting it with make menuconfig and comparing the differences.
Also, in the beginning of the thread, the post implies that ath10k-firmware-qca9984-ct is already compiled in the image ready to be removed should somebody want the non-ct firmware. However, when you build the image, it is not there. Luckily I had a non WIFI way into the router, and I could salvage the situation. Also, the diffconfig file does not build the kmod-ath10k (non-ct) that you can insert after.
I had to add:
CONFIG_PACKAGE_ath10k-firmware-qca9984-ct=y
to the diffconfig file to get the image to bring on the radios during various rebuilds I did. Another problem I encountered is that with the ct firmware, the 160MHz Width does stop the 5Ghz radio from coming on. I am currently with the kmod-ath10k-ct and with the non -ct firmware and then it worked.
I have also applied the NSS scaling suggested by @Ansuel (by manually editing the file) while keeping performance governor. It was stable the whole day, working, online meetings, but it just decided to reboot out of nowhere when the traffic load was very small.
The most impressive is the temperature drop, from an average of 59.5C in the past 7 days (running 19.07) down to 57C running 21.02 + NSS.
I don't game, so not chasing after bufferbloat, nor I need 1Gb LAN as I am still on cable ... very slow 
What I hoped was better WiFi performance but it seems that one of the laptops is capped at 400Mbits/s anyway, and a short test with a newer one showed an average speed of 650Mbits/s with iperf via WiFi - but this I did not test with the 19.07 stock to have something to compare to.
I would also like to understand more why it is said that WiFi cannot be totally offloaded to NSS, but is too early to start asking questions, I need to read some more first.
Thank you again
@ACwifidude @KONG
With KONG'S modified patch,my ea7500 keep working for 3days by ACwifidude's lastest nssmaster repo then dead (no wifi and pc can't get ip address from lan port)
thank you
Replace the entire first part of the diffconfig with this and it’ll pull in the defaults for the r7800 and build for the r7800 only (-ct is the default, you can specify if you want ath drivers):
# Use "make defconfig" to expand this to a full .config
CONFIG_TARGET_ipq806x=y
CONFIG_TARGET_ipq806x_generic=y
CONFIG_TARGET_ipq806x_generic_DEVICE_netgear_r7800=y
Glad it is working well for you. Let me know what you think about the wifi performance compared to your previous 19.07
You can flash back and forth so feel free to flash to 19.07, test, then flash back and test. Bet you’ll see a healthy difference. 
Today I start testing ea8500 with kong's new patch by ACwifidude's lastest nssmaster repo
Unfortunately, the router (g10) is very unstable. Restarts occur differently. Sometimes the roter works 4 days and sometimes 10 hours. There is no rule for this. The worst part is that the router disconnects wifi for a few seconds several times a day. My wife and I work remotely on various vpn and I cannot afford such disconnection. It is very tedious. So far I haven't figured out what can cause restart ...
Hello @ACwifidude, what is the consensus as to getting this stuff into the official OpenWRT master ? I ask because technically this stuff should make the router insanely faster in terms of routing traffic and the only way it will get more attention is if this stuff gets merged into master and more devices decide to test this feature. Sorry if I'm over-simplifying things.