I just tested some WRTs on my debricked router and sadly I experienced some major drawbacks.
So this is an N device and the maximum theoretical speed is 300 Mbps (+ LAN is 100 Mpbs). First I tested a few stock firmwares and the max bandwidth was ~93-94 Mbps. Since it was in AP mode I think it is an okay result.
Then I tried the actual stable OpenWRT 23.05.5. On LAN there is no noticeable issue. Usually I am sharing input devices (with Barrier) between two PCs and while I connected to this router it felt like never before. At this point I started to investigate what the hell going on. Even if the latency is better with OpenWRT it is laggy and the maximum wifi bandwidth is about ~16 Mbps. I thought it might be just a bad build, so I tested snapshot r27964-a0eafc3c77. And experienced the same exact issues. The devices are in a ~1 m radius on a desk. Distance must be not an issue. I also have another unit with Gargoyle 1.14 and it is performs the same way.
No. To simplify: I have two PCs on the same desk and sharing INPUT devices between both. And the first thing what I noticed after OpenWRT installation that the mouse tracking is laggy like hell because of wifi performance. Here are the details:
That is typical 20Mhz 2 antennas - total capacity 144Mbps -> you are getting half, which is expected.
You can add radio option
option acs_chan_bias '1:0.8 5:0.8 9:0.8 13:0.9'
And use auto channel before trying to force 40MHz
Rather strange - WPA3 uses same crypto offloaded in card as WPA2 , the problem is handshake only, not speed.
You mean it does not even use 40 Mhz even if it's set? I would like to use this router as an extender/repeater. I don't expect 300 Mbps of course, but it would be already a big improvement if it can reach the stock firmware's wifi performance.
Could you please tell me how can I add the mentioned option? I have some clues but far from an expert.
This is not exactly clear for me. But if you asking about the idle CPU usage it is normal. Also in my area there are only 2-3 other routers with weak signals and mostly not even overlapping.
Make pure WPA2 and pure WPA3 setup and re-measure. Some generations of e.g. Android had problems with mixed mode.
The difference should be in initial handshake being 1 or 2s long, not stream speed.
WPA3 has downgrade prevention, you need to clear connection to connect to same WPA2 name again.
Measure both cases (and for reference wired-nat-ed connection if possible)
Should be in the range of 70Mbps or 35Mbps depending on antenna count on the client.