Linux Should Soon Start Receiving "Make WiFi Fast" Improvements

[smitty3268]

What exactly is so bad about wifi? The last time I really had wifi problems was nearly 10 years ago when I still needed to use ndiswrapper.

As for audio, I agree - that needs improvement. ALSA is in desperate need for restructuring (configuring it manually is almost guaranteed a waste of time), and pulseaudio still feels incomplete, though at least it's very functional today.

This appears to be more about wireless access points (serving wifi) rather than your laptop using it.

The numbers they show in that blog look pretty good.

[starshipeleven]

What exactly is so bad about wifi? The last time I really had wifi problems was nearly 10 years ago when I still needed to use ndiswrapper.

linux (open drivers) wifi usually has shorter ranges than windows wifi on same hardware.

Also, most issues is about actually using the bandwith offered by n and ac wifi specs, something you notice only if you connect to a NAS over wifi, just using wifi for internet access does not show issues as that is tiny.

[starshipeleven]

The new wireless daemon will simply benefit from all these enhancements, correct?

https://phoronix.com/scan.php?page=n...ireless-Daemon

Yes, as these enhancement happen at driver level. Also wpa_supplicant does.

[starshipeleven]

linux (open drivers) wifi usually has shorter ranges than windows wifi on same hardware.

Also, most issues is about actually using the bandwith offered by n and ac wifi specs, something you notice only if you connect to a NAS over wifi, just using wifi for internet access does not show issues as that is tiny.

Also using VoIP, streaming movies at high-ish bitrates and whatever other traffic that needs good priority and low latency to not stutter horribly.

[starshipeleven]

unapproved post above

[torsionbar28]

Just to be clear, I'm not complaining about wifi getting better. I just don't understand how much better it could really be outside of implementing stuff like AQM or MU-MIMO. But, the article doesn't really specify if stuff like that is the primary reason for the improvements, in which case, most of us are not going to be affected.

Depends on the chipset vendor. Some work reasonably well, some work really well, some kind of sort of work, and some dont work at all. Wifi is one of those areas where you still have to be fairly choosy and do your homework before buying. I have Dell laptops where I've had to replace the included wifi card with a different one to get good results. The intel centrino 6200 a/g/n card works great, but the "standard" Dell one required me to build and install custom kernel modules, and another vendor's card worked with default drivers but had horrible range and bad performance.

Part of the issue is the lack of uptake in the business world, for Linux on desktops and laptops. After all, these are the types of devices that depend on wireless cards. Most of the big corporate kernel contributors focus on server functionality, since Linux rules the roost there. Outside of hobbyist home use cases, nobody puts a wifi card into a server.

[pheldens]

would love to try it on my 9k hostapd AP are there mainline patches?

[Creak]

My experience with Wi-Fi on Windows is that I've seen people losing their Wi-Fi signal over and over again, for no reason. They often have to delete the Wi-Fi network and add it again and prey for it to work... In other words: try to turn it off and on again

On Linux, I never had such a thing. Sometime an interruption but I still don't know if it's the Wi-Fi drivers or the Wi-Fi router or even my ISP. Anyhow it's too short to pay too much attention. So if Linux Wi-Fi drivers can be as reliable as they are today but in a more efficient way, I'm in!

[pal666]

Right... because the average Linux PC is in serious need for that feature (sarcasm).

for example, i have a policy of not buying hardware without mumimo

In terms of speed and reliability, I haven't seen any issues with that. Generally, wifi works just fine.

how did you measure that? those guys provide graphs of latency over slow wifi link under load. what are your numbers?

[pal666]

The new wireless daemon will simply benefit from all these enhancements, correct?

users will benefit from these enhancements, not daemons