autotldr@lemmings.world
on 12 Jun 01:35
nextcollapse
This is the best summary I could come up with:
The extensible scheduler “sched_ext” code has proven quite versatile for opening up better Linux gaming performance, more quickly prototyping new scheduler changes, Ubuntu/Canonical has been evaluating it for pursuing a more micro-kernel like design, and many other interesting approaches with it.
Torvalds feels the sched_ext code is ready enough and provides real value to the mainline Linux kernel.
This whole patchset was the major (private) discussion at last year’s kernel maintainer summit, and I don’t find any value in having the same discussion (whether off-list or as an actual event) at the upcoming maintainer summit one year later, so to make any kind of sane progress, my current plan is to merge this for 6.11.
I’ve never been a huge believer in trying to make everybody happy with code that is out of tree - we’re better off working together on it in-tree.
And using the “in order to accept this, some other thing has to be fixed first” argument doesn’t really work well either (and that has been discussed for over a decade at various maintainer summits).
So short of any last minute change of plans between now and mid-July when the Linux 6.11 merge window opens, looks for sched_ext in the next kernel cycle.
The original article contains 432 words, the summary contains 206 words. Saved 52%. I’m a bot and I’m open source!
ozymandias117@lemmy.world
on 12 Jun 02:29
nextcollapse
Wow! I didn’t expect sched_ext to be accepted based off historical precedent of not allowing multiple schedulers
I thought the focus would be on optimizing EEVDF now
This is going to be such an awesome addition. I saw an interview with one of the guys working on the project and it sounded so cool. I think the interview was on tech over tea with David Vernet. Great talk about the magic box thay is the Linux kernal.
So can I bang on the Intel 12th gen P-cores with the enterprise microcode to see if they are really fused? I promise to isolate the CPU set if I can pretty please have access to the full AVX instructions for use with llama.cpp. Pretty please.
bumpusoot@hexbear.net
on 12 Jun 08:33
nextcollapse
I didn’t realise Linus still wielded this kind of power. That’s kinda rad, I like it.
Evil_incarnate@lemm.ee
on 12 Jun 14:21
nextcollapse
The potential for distros optimized for specific tasks without needing to swap out entire kernels. A “gaming” focused scheduler probably looks different from a big data cruncher or a super multi-tasker server.
I like that he is being decisive about it. I wouldn’t be surprised to learn that the feature was only being delayed because of internal project politics or quirky policies that normally make sense, but don’t in this specific scenario.
threaded - newest
This is the best summary I could come up with:
The extensible scheduler “sched_ext” code has proven quite versatile for opening up better Linux gaming performance, more quickly prototyping new scheduler changes, Ubuntu/Canonical has been evaluating it for pursuing a more micro-kernel like design, and many other interesting approaches with it.
Torvalds feels the sched_ext code is ready enough and provides real value to the mainline Linux kernel.
This whole patchset was the major (private) discussion at last year’s kernel maintainer summit, and I don’t find any value in having the same discussion (whether off-list or as an actual event) at the upcoming maintainer summit one year later, so to make any kind of sane progress, my current plan is to merge this for 6.11.
I’ve never been a huge believer in trying to make everybody happy with code that is out of tree - we’re better off working together on it in-tree.
And using the “in order to accept this, some other thing has to be fixed first” argument doesn’t really work well either (and that has been discussed for over a decade at various maintainer summits).
So short of any last minute change of plans between now and mid-July when the Linux 6.11 merge window opens, looks for sched_ext in the next kernel cycle.
The original article contains 432 words, the summary contains 206 words. Saved 52%. I’m a bot and I’m open source!
Wow! I didn’t expect sched_ext to be accepted based off historical precedent of not allowing multiple schedulers
I thought the focus would be on optimizing EEVDF now
This is going to be such an awesome addition. I saw an interview with one of the guys working on the project and it sounded so cool. I think the interview was on tech over tea with David Vernet. Great talk about the magic box thay is the Linux kernal.
I believe this is the video: m.youtube.com/watch?v=Ta0imAIz31M
Thanks!
Here is an alternative Piped link(s):
https://m.piped.video/watch?v=Ta0imAIz31M
Piped is a privacy-respecting open-source alternative frontend to YouTube.
I’m open-source; check me out at GitHub.
So can I bang on the Intel 12th gen P-cores with the enterprise microcode to see if they are really fused? I promise to isolate the CPU set if I can pretty please have access to the full AVX instructions for use with llama.cpp. Pretty please.
I didn’t realise Linus still wielded this kind of power. That’s kinda rad, I like it.
It’s a heavy crown to wear.
Queue the “I’m still worthy” thor meme
Sounds good but what does sched_ext do and mean for the future?
Changing schedulers on-the-fly, depending on what you need to do on your machine.
Hasnt android (or at least custom roms) had this since forever?
I think you’re talking about CPU governors (conservative, ondemand, powersave etc.)
Oooh that’s cool
Holy shit that is one heck of a thing to do.
Does this mean anything for me as a consumer?
The potential for distros optimized for specific tasks without needing to swap out entire kernels. A “gaming” focused scheduler probably looks different from a big data cruncher or a super multi-tasker server.
Awesome. No idea what that means, but AWESOME!
Same. No idea what it means, but I like when Linus throws stuff.
I like that he is being decisive about it. I wouldn’t be surprised to learn that the feature was only being delayed because of internal project politics or quirky policies that normally make sense, but don’t in this specific scenario.
Finally, recompiling the kernel to change scheduler will be a thing of the past…