Instead of letting your anger explode into this thread why not provide constructive criticism? Posts like yours do no make the "problem" better. Except maybe for you, because you can vent some of your anger. If you want these problems with the 2.4 GHz WiFi and USB3 fixed tell me about it and ask if other people have the same experience as you do. Then, together, we can look into fixing them if that is within our reach or contact a developer about it. That is constructive and will make things better. OpenWRT uses open source drivers which are not feature complete but do run on newer versions of the Linux kernel with newer packages. Padavan and consorts use the closed source drivers and therefore are relegated to an outdated version of the Linux kernel and older packages. Pick your poison.
As stated in the OP: "There be dragons here!" However, over time, on a bumpy road, the dragons will shrink or even vanish due to developers chipping away at them in their free time. That is how open source projects work. An idea would be to do a crowdfunding action where we would hire NBD aka Felix Fietkau to work on the mt76 driver for a week. That might speed up things a bit. What do you guys think?
Hi labrok,
I saw you have a 868L. Dou you also use the openwrt on it ?
Is it rev A or B ?
kr,
Frank
I'm up for this idea! 
But to be honest i don't even have much problems with my DIR-860L at the moment. It's running stable and the Wifi performance is okay for me.
Some test results with r7944 running for 2d 1h 45m 46s
Gigabit wired (cat6/7 cables to the router - my usual connection for everything I care)
Test Complete. Summary Results:
[ ID] Interval Transfer Bandwidth
[ 4] 0.00-30.00 sec 2.95 GBytes 846 Mbits/sec sender
[ 4] 0.00-30.00 sec 2.95 GBytes 846 Mbits/sec receiver
CPU Utilization: local/sender 48.1% (6.2%u/41.9%s), remote/receiver 12.2% (1.4%u/10.9%s)
5GHz
Test Complete. Summary Results:
[ ID] Interval Transfer Bandwidth
[ 4] 0.00-30.00 sec 627 MBytes 175 Mbits/sec sender
[ 4] 0.00-30.00 sec 627 MBytes 175 Mbits/sec receiver
CPU Utilization: local/sender 13.5% (1.2%u/12.2%s), remote/receiver 1.4% (0.2%u/1.2%s)
2.4GHz
Test Complete. Summary Results:
[ ID] Interval Transfer Bandwidth
[ 4] 0.00-30.01 sec 239 MBytes 66.7 Mbits/sec sender
[ 4] 0.00-30.01 sec 239 MBytes 66.7 Mbits/sec receiver
CPU Utilization: local/sender 5.8% (0.7%u/5.1%s), remote/receiver 2.2% (0.4%u/1.8%s)
I'm perfectly fine with those speeds (as long as they are stable), probably with usb wifi dongle i can get double, but i don't need to.
iperf3 -V ...
Thanks, worked!
Nope, there are many custom builds in ddwrt and it also works fine in main DD-WRT firmware, in openwrt havent found anything! try the custom one from here https://app.box.com/s/b4v1s342ef2dpd52j02lqtuzi1oiipco/folder/38154548439. both work fine in both bands! The custom one uses older drivers for better combatibility, my opinion is that the old driver causes many TX/RX errors, its up to you which one you will chose, for the first flash try the one from the link i am giving you, there are reports that some routers brick randomly by the main DDWRT firmare! the main point of the guy who made the custom firmware is to avoid those bricks.Mine is revision C but all uses the same HW so dont worry about it!
There's been tons (and I mean it!) of commits in the mt76 driver repository. Are they rewriting large parts of it? An experimental build with the mt76 makefile pointing to the latest (and hopefully greatest) commit would be awesome for those people who are adventurous 
If it compiles at all in its current state that is.
Interesting! Been a while since I checked the repo. If I have the time I'll fire up a new build tonight/tomorrow.
My new job is keeping me busy so I don't have free time to tinker around.
Edit: waiting for my train on 22:00 so no build tonight 
Do you guys think this isn't a WIP (code refactoring), if i'm looking at the correct repo there are plenty of kernel panics reported?
I think they just added a new chip support for Mediatek MT7610U based USB Wi-Fi dongles and refactor their common functions, header files etc.
Correct, referred to as mt76x0 in the commits.
However, there are a few code cleanups to the mt76 driver, mostly memory allocation stuff, if I read the commits correctly.
please tell me that a non experimentall code is at the repos, the router already has become 4 years old and no stable driver has become available, tried DD-WRT also but it was absolutely failure, not even alpha testing code......5Ghz ridicolous 5 meters away from router and signal lost, Bandwidth.....hahahaha VHT80 not even to be mentioned i wonder why they are still having such a ridicolous firmware in their repos. The only solution is your builds or i will put the router in a box and forget it!
Do you check the logs or something, i'm not big on using wifi at all but don't have such issues on the mobile phones. There were couple of builds which had problems (totally stuck router after some 5Ghz traffic and some 2.4Ghz issues previously) but seems stable right now. Also ran some iperf3 for a while after latest flash from the laptop and it was consistent. For the bandwidth it really depends on the wifi cards in your devices.
Another option which you can consider is returning it to stock firmware (which after the redesign of the GUI lacks a lot of the features which were there before) and grab an Orange Pi one/lite/pc etc for like 20$, put armbian and run the things you wanted on the router.
i am talking about dd-wrt builds for DIR-860L rev B not this one here (which is the only working), last one on DD-WRT was a total disaster, tried it on 860 and the issues made the router useless, if i cannot use the 5Ghz radio in bigger than 5 m distance this is not considering as "stable"
Strange! Since they use the closed source drivers, you would expect things to be more stable. A build with the closed source drivers is still on my to-do list but my work (teaching) keeps me busy. Would be nice to compare the closed source drivers to the opensource drivers and both to DD-WRT.
@All, a new build is compiling atm. Tested on a MIR3G and running like a champ. One change I made, is disabling USB because it can interfere with wifi. Xiaomi does not enable USB3 but they did enable USB2 for that reason. USB throughput is almost identical so not as big of a deal as it appears. I disabled it completely to see how the wireless part of the router performs without it activated.
For people who want to test it you can download r8064 here. Let me know what your experiences with this build are. Keep in mind that this build does not support USB storage!
for me the same issue with 5Ghz was 2 years ago with the first Dir 860L builds, same today, this shows that nobody really cares about this router at DD-WRT....there is a Padavan build which seems very good, but its a bit tricky on installation nad vulnerable to bricks by accident, maybe you should take a look at padavan `s firmware if it is possible .
On r8064 wifi 5g is really unstable on both my phone and laptop, connection drops every few minutes.
r7682 was a little bit better but also I had some connection drops.
For me the latest stable version is the 18.06.1 official release, using it I am stable wifi 5g (even if the changelog says that isn't!!).
Since I love bartvz releases I am waiting for new releases in order to be back with my favorite optimized openwrt firmware
On r8064, did the syslog reveal anything? Something like: "disconnected due to excessive missing ACKs"?
Are other people facing similar issues?