Optimized build for IPQ40xx devices

@NoTengoBattery Great, Let me know as there is a build to test

@uniqe13 would you be able to post the calibration file. I could not able to generate this.
I would like to try and see if this improves the stability of 2.4 GHz WiFi. Do you also use -ct firmware?

File board calibration for ea6350v3 with DK06... 2G+5G settings is here

Firmware of wifi non -ct version.
Kmod-ath10k but may be -ct version also.

How is it installed since the wifi in the 2.4 band is very unstable in my ea 6350 V3?

Grateful with this calibration, would the problems of the Wi-Fi 2.4 network be solved?

This calibration solves the coverage problem, the speed in 2.4G is better but there are still problems.

Copy file board-2.bin to locate /lib/firmware/ath10k/QCA4019/hw1.0 in router.

This calibration works fairly well for me, but there are others you can try that may work better.

Edit: For better performance in 2.4G install non -ct firmware.

Hello all. Just a heads up that I was unable to get my upload speeds working with this build when connecting to an Enterprise networking using PEAP / EAP-MSCHAPv2 after upgrading to full wpad-wolfssl. Here's a post with my issues. I ended up going with the stock build + wpad-wolfssl and got it working.

Thanks. It's the most stable and coverage of the calibration file. Streaming 2 IP camera using rtsp to Nas and it has not drop the connection for 1 week. Bandwidth not the greatest.

Next tests finished- TCP Congetion Control algorithm in order from most efficient for my test environment (2.4G, 10 meters)....
TCP YeAH>TCP NV> TCP Hybla> TCP Veno>TCP BBR.... Edit: But it is maximal 2-5% difference .
There is a C2TCP implementation for cellular environments, but implementing it for OpenWRT is beyond me :).

I've added the Minstrel-Blues patches for testing, but I don't see any changes compared to Minstrel-HT.

Generally there are still problems with lower transfer rates in 2.4G (even near router).

Many many thanks for this great work.

I have reported an issue here regarding 2 packages found here generic/package v3.0.1 that seem not matching kernel.
Note that rest of packages needed for wireguard are ok.

oh I didn't see there was already a post on wireguard

hello members, I come to show you my experience of what happens with the firmware in question, I have 500/150, it happens that I used the stock firmware and could reach the speed of 50/50 on 2.4ghz, but after the update I can't do more than 10/30 , strangely the upload is not affected, could someone help me?

hi, I read somewhere that it has to do with incorrect or rudimentary handling of TX TSO and the like in openwrt- there are some patches but they don't add all the functions (supported by ethtool), and what they add works ambiguously and still has bugs. Due to these deficiencies, there are bandwidth drops, sometimes only on 2.4 but also on 5GHz and mainly on the TX line (this is a download link for wireless connectivity).
The problem may be in what I described above, it may also be something else, but it occurs, as far as I know, on a large number of qualcomm devices - also older and theoretically already known (e.g. Tp-link 1043nd v1, v3). Also on ipq4018 (ea6350v3).

how can i solve the problem? I tried to use test calibration but I don't know how to use it because English is not my native language and I found it difficult to use it, someone would qualify

this problem will not be solved by calibration ... you have to wait for new openwrt releases because it is rather a programming problem. And this is not simple to fixed.

PS. English im my not native language also, and I think my his level of composure is bad. :wink:

but the firmware creator can't solve this problem? Do a kind of hack in the software and extract the files that make it work correctly, this router is very powerful because I can shape 500/150 with sqm and this speed on 2.4ghz wifi annoys me.

I think I've found a LuCI bug in v2.4 and v3.0.1.

LuCI > Status > Realtime Graphs -> Connections tab

Basically it is not reporting all connections.

I often use this tab to monitor how many MB has been transferred to/from an external IP on other Openwrt router devices running LEDE 17 to OpenWrt 21

The easiest way to reproduce the bug is to use a streaming video service (eg. I use BBC iplayer in the UK), and then check the above tab. No connection seems to be reported, with incrementing data usage for my device streaming from BBC.

I first noticed it on v2.4 (running uninterrupted for 36 days so far. Wifi disabled). Only additional package installed is SQM (or it may have been DDNS). Operating in normal 'router' mode.

I can reproduce problem in v3.0.1 out of the box with no changes to config. No extra packages installed. Operating in normal 'router' mode.

Update:
The stats displayed on Connections tab behaves differently with:-

EA6350 v3 - with 19.07.8. no config changes, no extra packages.
Connection stats working OK

After nlbwmon is loaded, connection stats are reset every 25-30 secs.

HH5a with 21.02.1
Connection stats working OK

After nlbwmon is loaded, connection stats are reset every 25-30 secs.

Hey @NoTengoBattery, I've seen you're doing a great work.
I'm currently using the "stock" OpenWrt build (21.02.2) on a FRITZ!Box 7530 as the main router (pppoe + nat for wan) with a 1Gbps/300Mbps FTTH.
With the stock firmware (AVM ones) it was doing full speed consistently, with OpenWrt is doing 150/200Mbps DL and 200/250Mbps UL without NAT SW offloading and 400 DL 250 UL with it enabled. Irqbalance made no difference whatsoever, in fact only the first core is always used.
Could one of your builds improve the speeds? Would you have the time and the will to try and do a build for this device (I'm obviously willing to test it)?
Thank you

This patch: mac80211 adjust QCA9561 PA bias
made me wonder if poor stock OpenWrt ipq4018 2.4 GHz EA6350v3 radio performance could be similarly addressed without calibration file changes?

It might need something similar to

greetings, recently I realized that there are people performing both dialing pppoe and dhcp, only a single core is used, this somehow limits the performance of the router, because even though it is strong in single thread it would never make a pppoe gigabit connection like stock firmware, 2.4ghz performance is really very complicated this router has incredible potential, but very difficult to program.