I think it runs like a bat on fire.
But I am also a bit confused!? What specific performance issue on the 5 GHz do we talk about?
I have been around in the forum a while and I can’t really remember some big performance issue treads in the forum.
AFAIK nothing has changed. The problem lies with the ath10k and ath10k-ct drivers, which I have no knowledge of, nor time to investigate. The radio on the EAP245v3 is a rarely used model, so nobody seems to be bothered enough by the fact it doesn't work perfectly (it's certainly good enough for me). If you care about getting maximum wireless performance, stick to stock. OpenWrt provides other benefits.
@flygarn12 Like @svanheule said, this device seems to have driver issues, which don't allow EAP245 to reach it's maximum throughput (~900 Mbps download) on 5Ghz, when using OpenWRT.
@flygarn12 Are you able to reach ~900 Mbps download with your EAP245? If so, there's maybe some unknown quirk, maybe related to the region variants (btw, I'm using an EU-European model).
@svanheule Thank you very much for the status update and thank you very much for your work.
It's really unfortunate, because this is the perfect budget pure AP (cheap, fast, good coverage and all the main components are Qualcomm).
Unfortunately, I will keep using the stock firmware for the time being (which is fine for my needs), but I will switch as soon as this ends up being fixed.
Being a fairly popular device (I think), I hope someone is able to fix the driver issues in the near future (I wish I could help with this, but it's way out of my league).
Btw @svanheule is there some place to keep me updated about this issue, aside from testing it myself?
Thank you very much @flygarn12 and @svanheule for your help.
This used to be a good thing with ath9k, but the situation has worsenend with ath10k (11ac radios).
IIRC there was a ticket filed with the ath10k-ct driver, but you would have to look it up. Don't get your hopes up though. People have reported throughput issues on this device nearly since the day it was supported.
Really? I thought Qualcomm was the best for open source drivers.
Is that not the case anymore?
I will look it up, thank you.
But will it be absolutely impossible to get the stock throughput on OpenWRT (due to lack of proper drivers)? Or is just the case that no one with enough/proper skills got interested in this issue?
Again, thank you very much @svanheule for taking your time to answer my questions,
Well to be honest, 900Mbps per client over wifi…I don’t have a internet line to fill it up with anyway. So I haven’t even tried.
From the beginning my ISP used to have free floating outgoing line and then I could get around 800Mbps output over wifi before a rebuilt the network with a vlan trunk. But now they have regulated the speed on both ways.
But it doesn’t matter anyway. I run the internet line on a Gbit duplex trunk connection inside the network so like 500Mbps is kind of full speed anyway.
Since my new neighbors recently moved in the wifi air got really crowded so now I down 20MHz channels.
But I never bought it for that speed anyway. I wanted multi VLAN, multi SSID and WPA3 and business class hardware to get rid of all this home devices crap.
Unfortunately, for me, maximum throughput is important.
For your use case is awesome, because this device meets all your needs.
When I bought my device (more than a year ago), I had an eye on EAP225 (which seemed/seems better supported on OpenWRT), but it was slower than EAP245 and seemed more prone to hardware failure, from the reviews I read. I also had an eye on the Ubiquity equivalent (AC Pro), but it was considerably more expensive (about 50% more expensive), for about the same performance (and was not always in stock).
WPA3 is the most notorious absence, compared to the stock firmware and one the most compelling reasons to switch.
Multi VLAN and multi SSID are present on the stock firmware (I use them both with success), albeit I think there is an hard limit to the amount you can have/create. I don't now if it's the case with OpenWRT.
As I mentioned, for my use case, the stock firmware is fine (much better and cleaner than other tp-link products, at least), but I would switch right away, if possible, to OpenWRT, because the community is awesome and the support is unmatched (and is open source, of course 
).
1 Like
https://forum.openwrt.org/t/tp-link-eap245-v3-stuck-at-40mhz-on-5g/96626/8?u=flygarn12
Have you tried changing the driver as this instruction mention?
1 Like
Thank you very much for point that out, the result looks very promising 
As I said previously, my device is being used constantly and intensively during the day, so I have my concerns about making sudden and drastic changes that may lead to extended downtime (the fact that is not dead easy and safe to go back stock firmware is also a big concern of mine).
If this driver change is the key to improve the device performance, I may give it ago, when I find a suitable time to test it.
Have you test it on your system(s)?
1 Like
Nopp, I just saw the post this morning.
Will you give a go in the coming days?
@svanheule Also, is there any problem in changing to different drivers? (stability wise and for future OpenWRT updates - do I need to have special care with future driver updates? Or are these non CT drivers automatically update with every system upgrade?)
Again, thank you very much.
I'm really glad you hijacked my original post (and let's say quite silly noob question).
This is really useful information.
I'd like the possibility to keep the stock firmware, at least while reorganizing my network.
And I need to pass multi-SSID VLANs to the router; I don't know many APs in the same price that allow this out of the box.
I've read that EAP225v3 has a better known hardware but it's indeed slower and older.
The EAP245 firmware fix could be a game changer.
TP-Link RE650 could also be a good candidate: flashing OpenWRT seems really easy.
Where I live it also costs quite similar.
You can submit a patch to change the default driver, if you want. You should argument why you want to make this change (throughput issues), since there are other differences in features provided by the ath10k and ath10-ct drivers. I don't know exactly what the differences are though, so you would have to look that up, or ask people more familiar with these drivers.
1 Like
We all start as noobs (with OpenWRT, I'm one too)
The stock firmware is decent and is dead easy to set these and work really well. The main problem (I think) is that there is an hard limit to the amount of SSIDs and VLANs you can have/create.
EAP245 is a very nice piece of hardware and, for me, at the time, ticked all the right boxes.
Having OpenWRT support is the cherry on the top.
Also, is not that easy (and cheap) to find a pure AP with the right specs and that also supports OpenWRT (at least where I am).
I will try to find what are the differences and what are the consequences of this change.
Is it common to have this driver variation/swapping during the hardware's lifetime on OpenWRT?
Or, usually, a device has a specific driver on OpenWRT and sticks to it to the end of hardware support?
Also, if I make this driver change, do I need to update the driver manually or is it automatically updated with every update?
Sorry if I'm making stupid questions, but all this is new and fascinating to me.
Again, thank you for taking the time to answer my questions.
I doubt it, I am working fully with the dgs-1210 now.
Have you tried searching this forum on the driver name to maybe see any background info to it.
Drivers do change both to good and bad over time.
https://forum.openwrt.org/t/why-the-switch-to-unstable-ath10k-ct/27258?u=flygarn12
But just google these drivers difference returns a lot of problems with the -ct version it seem.
1 Like
Yeah problems with -ct version seem very common like here:
But it is always device specific.
One thing I notice but without any scientific proof but the -ct driver was as I understand implemented around 2018 I think and many problems seems to be from 2021?
So was there a change in the driver in that time frame?
I dug a little bit about the ct vs non-ct drivers.
If I read correctly, the ct driver is a fork of the non-ct (mainline) driver, maintained by a single person.
Also, the ct driver is shared by a bunch of different devices, each with their own uniqueness.
What may happen with time is that changes that fix one (or more) devices may break others, which may explain why some devices improve with time and others get worse.
I may be completely wrong, though.
Meanwhile, @adamcstephens results for the non-ct driver don't seem to be very encouraging... (the ct driver had 2.4Ghz tier speeds, which is ridiculous / the non-ct had a significant improvements, but, apparently, it still pales in comparison with the stock firmware).
1 Like