I’m getting way lower network bandwidth than expected. Before I was able to saturate my internet bandwidth, now it can only do around 95~ in local. I was using fedora arm before and it was fine. I tested again with fedora arm (just replaced the sdcard with fedora one) and there was no issue.
Please try to use channel=0 or remove the line entirely. It then chooses the least crowded channel automatically. I aim to remove it with next DietPi release from our default config when adding 5 GHz support to our hotspot options.
I see you enabled WiFi 4 and WiFi 5 + QoS.
If I understand it correctly, ht_capab enables and fine-tunes 40 MHz wide channels. As long as you are not in a very crowded WiFi area, I’d remove this and trust in defaults being sane.
Not sure what require_ht and require_vht is?
Require stations to support HT/VHT PHY (reject association if they do not)
Ah found it, “HT” == “High Throughput” and basically means WiFi 4, “VHT” == “Very High Throughput” == WiFi 5. WiFi 6 seems to not add something to that, but WiFi 7 adds “EHT” == “Extremely High Throughput”.
So you are enforcing WiFi 4 and WiFi 5 support with these two directives. I assume it means the clients must support those, not the AP. And then you disabled ieee80211d and ieee80211h which are both disabled by default.
Did you compare the hostapd.conf with the one used in Fedora? It is a standard software and should be mostly the same on both distros, probably a slightly newer version on Fedora. Does Fedora also use the RPi kernel sources, or mainline kernel?
I tried removing some options from hostapd and tried many combinations, they dont seem to help. I also tried channel=0, which didnt help.
Did you compare the hostapd.conf with the one used in Fedora? It is a standard software and should be mostly the same on both distros, probably a slightly newer version on Fedora. Does Fedora also use the RPi kernel sources, or mainline kernel?
I’m not sure which kernel fedora uses it seems to be mainline since I couldnt get cpu temperatures on that kernel. hostapd config is the same.
Okay, the firmware version is as intended the one from the RPi instead of the one from Debian, so that should be fine.
I think next we should find someone else to replicate it. And it would be great to test the same with RPi OS.
Ah and we should be able to install Debian’s mainline kernel build, the linux-image-arm64 package. An initramfs will be required and the /boot/config.txt adjusted to load the new kernel image, dtb and initramfs. I hope it works well with the bootloader from the RPi repo.
actually, we need to find out differences between our installation and fedora. If you have time, it might be helpful if you could check with plain RPI OS, running same kernel and configs.
Just because the bitrate says 433Mbps, do not expect such speeds. It is theoretical speed, not achievable in reality. Even access points with better radio can barely get half of what they advertise.
On top of that, the wifi of the Raspberry is nothing fancy and I would not consider it a reliable source for speedtests.
Maybe you’ll be able to get a bit higher with channel and distance optimization, but eventually not so much.
Yes you are right, but I was getting consistent saturation of my internet bandwidth on speedtests, when on dietpi I’m only getting 92mbps~. I’m supposed to get 110mbps, which is still achievable and reproducible when I just plug in the fedora arm sdcard.
here are iperf3 results from fedora arm on the pi:
