Features You Won't Find In The Linux 3.9 Kernel

[phoronix]

Phoronix: Features You Won't Find In The Linux 3.9 Kernel

While there are many interesting features to the Linux 3.9 kernel, there is some functionality you will not find yet within the mainline Linux kernel...

http://www.phoronix.com/vr.php?view=MTMyMzQ

[duby229]

That article is a bit disingenuous. Some of those features arent going to make it it into -any- kernel ever.

[Linuxhippy]

So, if those features are not in 3.9, whats the reason writing about?
Seems you have a hard time meeting some secret goals of posting frequency you are trying to reach.

[zanny]

That article is a bit disingenuous. Some of those features arent going to make it it into -any- kernel ever.

Back in 2004 a lot of people would have said no one would have changed the O(1) scheduler, but then popular opinion turned against it.

[Serafean]

Among features that are being actively blocked from inclusion you missed tuxonice. But this article is a pretty nice write-up nice out-of-tree features.
Why mention not publically worked on stuff (UVD) is beyond me...

[BO$$]

Why isn't BFS in the mainline already? I have heard a lot of good things about it and that in some situations it's better than CFS? Why should he push for it to get it in the mainline? Shouldn't Torvalds add it himself if it's good for linux?

[birdie]

BFS won't be included unless someone reworks it to make it scalable. In its current state it doesn't scale well beyond 8 CPUs/cores.

Reliable Nouveau Re-Clocking is the most sought feature on this list.

[uid313]

Sad to see reliable re-clicking support for Nouveau isn't on the list.
Personally, I would really love to see that!

Lack of AMD UVD support is something that harms AMD, and in the coming months I might buy a new computer, and then AMD won't even be an option.
Not gonna buy something thats not porno-compatible.

Personally, I would like to see further ARM multi-platform work.
Further improvements to the work that was introduced in version 3.7 of the Linux kernel.

[BO$$]

BFS won't be included unless someone reworks it to make it scalable. In its current state it doesn't scale well beyond 8 CPUs/cores.

Reliable Nouveau Re-Clocking is the most sought feature on this list.

Can't they put an if (numCores < 8) { useBFS(); } else { useCFS(); } ?

[RahulSundaram]

Can't they put an if (numCores < 8) { useBFS(); } else { useCFS(); } ?

They could but they wouldn't want to. Linus has always emphasized that schedulers need to work well across the board to be acceptable. While 8 cores might seem fairly big now, mobile phones will be running that in a few years. None of the mainstream distributions will be interested in a scheduler that isn't generic but only works for some configurations.