I have a laptop with a RTL8821CE wifi chip and I'm getting a maximum of 20 Mbps no matter what.
Another laptop with an older intel chip gets 300 Mbps. My samsung smartphone gets 300 Mbps.
Wired I get the full 1 Gbps.
Yesterday I connected my laptop (the RTL8821CE one) to a public hotspot and was getting 150 Mbps.
I tried playing with TX powers on both the router and the laptop : no change.
I tried playing with the channel widths : no change.
I tried playing with different channels : no change.
I tried using 2.4 GHz : no change.
Tried turning off power management on the laptop : no change.

For example I tried setting TX power to 5 dBm on both the laptop (using iwconfig) and the router, setting channel width to 20 MHz on a 5 GHz band in A mode (no N, no AC, no AX), only 1 RX chain and only 1 TX chain, moving the laptop around so that I'd get a receive power of about -50 dBm on both the router and the laptop. This should give the most compatible speed reading right? The connection is marked as 54 Mbps. The phone gets a nice stable 25 Mbps down and 25 Mbps up. The laptop with the realtek chip 14 Mbps down and 16 Mbps up. The laptop with the intel chip got 22 Mbps down and 21 Mbps up and it was further away at -65 dBm.


Just puzzling.

I tried using 2.4 GHz

Unlike ax, ac isn't defined for both 2.4 and 5GHz. You would have fallen down to "n" at best. If it fell down to "g" instead, that would explain your speeds.

Assuming both radios are using the same SSID, I'd try turning on FT mode on them both, giving the laptop its best chance to use its ac capabilities on the 5GHz bands. Otherwise, it is likely to decide the 2.4 GHz band is more powerful, thus a "better" choice.

If that doesn't help, post your wifi configuration. Let's work from data, not guesses, okay?

I tried using 2.4 GHz

Unlike ax, ac isn't defined for both 2.4 and 5GHz. You would have fallen down to "n" at best. If it fell down to "g" instead, that would explain your speeds.

Assuming both radios are using the same SSID, I'd try turning on FT mode on them both, giving the laptop its best chance to use its ac capabilities on the 5GHz bands. Otherwise, it is likely to decide the 2.4 GHz band is more powerful, thus a "better" choice.

If that doesn't help, post your wifi configuration. Let's work from data, not guesses, okay?

I tried with 2.4 GHz completely disabled.
With 2.4 GHz completely disabled I tried different modes : A, A/N, AC, AX : no difference whatsoever, can't break the 20 Mbps barrier.
With 2.4 GHz still disabled, I tried different combinations of channel widths : A 20 MHz, A 20/40 MHz, A/N 20 MHz, A/N 20/40 MHz, AC 20 MHz, AC 20/40 MHz, AC 20/40/80 MHz : no difference, still won't break the 20 Mbps barrier.
With 2.4 GHz still disabled, I tried different combinations of modes, widths, TX power settings, chains (tried only chain 0 for both RX and TX, tried only chain 1 for both RX and TX, tried chain 0 for RX and chain 1 for TX, then chain 1 for RX and chain 0 for TX, also using both chains for RX and TX), nothing gets me above 20 Mbps on this specific laptop.
Another laptop (intel wifi chip) gets 300 Mbps with the same hAP ax3, in the same exact position, that this specific laptop (realtek wifi chip) gets 20 Mbps, and the same laptop (realtek wifi chip) gets 150 Mbps on a public hotspot.

Put the realtek in the real garbage.............

I repeat: show your configuration. At minimum, the output of:


For comparison, here's a boiled-down version of my ax³ config, which gets me near a gigabit right on top of the router with an ax client, and 200-300 Mbit/sec a few rooms away.


If yours is substantially the same, then my suspicion is that we're down to implementation compatibility issues. "Realtek" is a bad sign, and I'm getting the impression that you're talking about a Linux host atop that, where Realtek puts in even less effort than normal.

I repeat: show your configuration.

Here's my configuration right now:
Output from iwconfig:

More complete:

One of the reasons I suggested using an /export command instead of /print is that it suppresses sensitive info like your WiFi password by default. I've edited that out of your postings above, but you can't count on us moderators to backstop you like that every single time.

Other than local details like that, I see no meaningful difference between your configuration and mine. You could let some of the values go to defaults instead of the local overrides you have — e.g. I see no reason to override the default mobility domain value on a simple home WiFi gateway setup — but I don't see that that would change anything.

You wanted a new 802.11ax WiFi dongle anyway, right?

One of the reasons I suggested using an /export command instead of /print is that it suppresses sensitive info like your WiFi password by default. I've edited that out of your postings above, but you can't count on us moderators to backstop you like that every single time.

Other than local details like that, I see no meaningful difference between your configuration and mine. You could let some of the values go to defaults instead of the local overrides you have — e.g. I see no reason to override the default mobility domain value on a simple home WiFi gateway setup — but I don't see that that would change anything.

You wanted a new 802.11ax WiFi dongle anyway, right?

My laptop has one single USB C port and I can't find a small USB C WiFi 6 dongle. I don't want to use an adapter. I want something small and unobtrusive, that won't break if I apply some force.
Buying another laptop... hard to find one that has WiFi 6, lightweight, touchscreen, works well with linux, very long battery life (I get 9 to 12 hours out of this cheap $350 Asus TP401, with a puny 2 core CPU and 4 GB of RAM), and cheap enough that I don't worry of it being stolen or damaged. With Devuan this laptop works surprisingly well. I have 15 tabs open in firefox right now and RAM usage is 2.72 GB of 3.65 GB available. Just opened the thunderbird email client in parallel and it's 3.14 GB/3.65 GB. The thing has no fans and makes no noise.
Works fine with non-MikroTik wifi. Other devices work fine with the MikroTik AP.
Maybe I'll open it up and see if I can replace the wifi card.

WiFi 6, lightweight, touchscreen, works well with linux, very long battery life, and cheap enough that I don't worry of it being stolen or damaged.

Sounds like a mid-range Chromebook. Mine manages to push 180 Mbit/sec to the Internet through the ax³ from a few rooms away. It's a 2021 model, so one hopes a current model will do better.

Ironically, this one appears to have a Realtek WiFi 6 radio inside. Difference being, Google has a big, weighted bat to swing around, allowing it to get Linux compatibility issues solved.

WiFi 6, lightweight, touchscreen, works well with linux, very long battery life, and cheap enough that I don't worry of it being stolen or damaged.

Sounds like a mid-range Chromebook. Mine manages to push 180 Mbit/sec to the Internet through the ax³ from a few rooms away. It's a 2021 model, so one hopes a current model will do better.

Ironically, this one appears to have a Realtek WiFi 6 radio inside. Difference being, Google has a big, weighted bat to swing around, allowing it to get Linux compatibility issues solved.

I tried booting windows 10 from a usb disk to see what happens. Got 170 Mbps.

So title should be changed to "Chromebook with RTL8821CE = bad wifi speed but works normal using Windows"

This is no Mikrotik problem.

So title should be changed to "Chromebook with RTL8821CE = bad wifi speed but works normal using Windows"

This is no Mikrotik problem.

Well, the OP reported that even on his Linux, but on non-Mikrotik based wi-fi it works as well fast enough, so there must be some Mikrotik specific setting that is not compatible with the laptop setup/drivers/whatever or viceversa something in the laptop setup/drivers/whatever that is not compatible with Mikrotik (default) settings.

I have client devices not behaving on non-MT APs but they work nicely on MT. Where should I complain ?
You can never cover all bases for every possible combination of client devices and access points.

Just have a look at all the Apple-related wifi problems.
Everyone and everything is at fault except Apple ? Yeah right ...

What OP could try is to set everything on that AX3 as minimal as possible (only WPA2 maybe even WPA, no encryption, ...) and then see what happens.
Also separate SSID for 5GHz channel to see if his device connects then to the proper radio.
Yet it does all work nicely with his smartphone or older laptop, yes ?

So IMHO all this is not an AP problem. It's a client problem (client = combination of HW and OS).

Put the realtek in the real garbage.............

I was just trying to remember what gave me so much trouble, just dug it out to take a pic.....

@holvoeth
I don't think anyone is attempting to blame Mikrotik for the issue, and actually it is rather pointless to blame anyone, there is a problem and this problem should be solved, one way or the other.
If the device (or the Os or the driver) is incompatible with *something* that Mikrotik does (possibly with 100% compliance with any and all RFC's, technical norms and whatnot), the point is to find that *something* and a way to change it so that it works.

It is - as I see it - not unlike the approach used (from what we can see on TV) by the NTSB when an airplane accident happens, the scope is not primarily that of finding a culprit, but rather to find a way to avoid the same accident to happen again.

I know where you're going to and still my position is: it depends.

If that workaround implies you have to lower standards in such a way that you simply can not pass anymore between floor and bar, then one should consider the option to simply let go of the offending devices.
You can not keep supporting everything or at a certain point you will not be able to move on.

E.g. I had with an installation on a specific site of a client ONE single device not allowing me to use WPA2 (wifi card of a paint color mixing machine).
Even allowing both options WPA/WPA2 was enough to make it choke. Only WPA was accepted as valid option, nothing else.
I replaced the wifi dongle on that PC and that was it.

Here is some more information if it helps debug the issue:
OS: Devuan GNU/Linux 4 (chimaera) x86_64
Host: VivoBook_ASUSLaptop TP401MAR_TP401MA 1.0
Kernel: 5.10.0-27-amd64
CPU: Intel Celeron N4020 (2) @ 2.800GHz
GPU: Intel GeminiLake [UHD Graphics 600]

It's using in kernel driver rtw88_8821ce.

Wifi works fine with non MikroTik hardware. Also works fine under Windows with MikroTik hardware.
I have not tested other linux distributions but if you want me to test it's not a problem, as long as I can boot a live ISO.

Linux kernel 5.10 ? That's from ... what, end 2020 ?

There are updates for that RTL8821CE driver going to januari 2024.
I would have a look there first.
Update your drivers.

True story:
new laptop: windows 11 Pro, (also Realtek) ethernet card with normal port and USB.
Couldn't get over 840Mb using ethernet connecting to CSS610 or AX3 (and then iperf server on NAS). Not via ethernet port, not via USB-C port to docking station. I wasn't able to understand why.
Updated drivers directly from supplier and it's humming nicely at 950-ish now.

Yep , I concur that "it depends", but "it depends" is different from the initial (perceived) "not a Mikrotik issue, go away".

Let's see if we can find on what exactly "it depends".

Obviously not at the same amplitude level of the (known) Apple issues, but an Asus Vivobook is not a rare device (though likely right now 99% of them are running WIndows and don't have this issue) but it looks like one that will likely be discarded by Windows users and that will fall in the hands of Linux users, not a niche, one-user-only-has-it device, and probably the problem is connected to the wifi card and/or its driver, so it likely extends to all notebooks with that card running Linux.
Replacing the card in a notebook, besides the cost of the new card, can be - depending on the model - a difficult task, if it was a desktop PC I would have suggested to try a new nic first thing, with laptops is not so easy.

@ansky
You should provide a complete configuration export, follow this:
viewtopic.php?t=203686#p1051720
there is still the possibility that the problem lies in some obscure setting in Mikrotik that some experienced user of the forum can spot and that can be changed without compromising your network security and without the sky falling on our heads (tomorrow).

I don't think that a different distro, likely using the same rtw88_8821ce driver, would behave differently, no idea if any other Linux (or - say - BSD) coming in a .iso bootable/volatile form BUT using a different kernel driver actually exists.
Very likely it won't make any difference, but for what it costs trying booting a non-debian derived .iso is something to try.

To recap, possible causes may be:
1) some "default" or "user modified" settings on the Mikrotik side <- if you provide your configuration we can check if anything is blatantly "wrong" in it.
2) some "default" or "user modified" settings on the Linux side <- you should know if you have changed anything and/or you can try resetting to default
3) some "peculiar" setting/behaviour of the driver <- in which case maybe there are other documented cases on Linux dedicated boards/mailing lists/whatever
4) some "peculiar" behaviour of the Asus Vivobook hardware with that driver

If #1 then Windows (or its driver) has a way to workaround it by default
If #2 then all distros will have the same settings and will fail in the same way (so it would be more likely to find similar documented cases)
If #3 it would still seem to me improbable that noone else has already experienced the same issue, Mikrotik devices may be not the most common around, but they are not niche, and as well that Realtek card is not a home made single specimen, but if you are running an older driver I would try latest first thing
#4 seems like the most probable (and also the most difficult to pinpoint/troubleshoot, unfortunately)

P.S.: Check this:
https://forums.linuxmint.com/viewtopic.php?t=394684

I don't think that a different distro, likely using the same rtw88_8821ce driver, ...

A different distro should then also use that same 2020 driver in order to compare apples to apples (yeah, used that one on purpose ).
Let's have that driver first updated to current version and then we can see how to proceed.

https://www.amazon.ca/s?k=wifi+pcie+car ... _sb_noss_2

OR

https://www.amazon.ca/usb-wireless-adap ... ss+adapter

P.S.: Check this:
https://forums.linuxmint.com/viewtopic.php?t=394684

That did it. Using driver rtl8821ce instead of rtw88_8821ce.
I got 186 Mbps download and 67 Mbps upload. Still not the 300 Mbps that my other devices are getting but it's acceptable.
Why the problem presents itself only with MikroTik AP combined with the rtw88_8821ce driver is still an open question. Although I can't exclude that the problem may present itself with other routers. I haven't tested all the routers out there.

Edit: using TX power 10 dBm (on the laptop) I got 180 Mbps download and 97 Mbps upload. My WAN is 1 Gbps download and 100 Mbps upload.

Edit: using TX power 10 dBm (on the laptop) I got 180 Mbps download and 97 Mbps upload. My WAN is 1 Gbps download and 100 Mbps upload.

So maybe the 180-190 is like the max speed available on that notebook?

I have not understood if on that notebook you are dual booting Windows and Linux, if not testing another notebook (possibly more powerful/with better Wi-FI) running Windows can tell us very little, besides the fact that in the same spot a faster connection is available.

In theory, the RTL8821CE should be a 433 Mbps device, which in practice should max out (in lucky days) around 2/3 of that so - maybe 300 Mbps.

That card/chip (and/or its drivers) are often reported as slowish and/or fluctuating even on Windows (particularly - it seems - on HP laptops).