Linux 5.0-ck1 Kernel Rolls Out With MuQSS 0.190 Scheduler

[phoronix]

Phoronix: Linux 5.0-ck1 Kernel Rolls Out With MuQSS 0.190 Scheduler

One week has passed since the official debut of Linux 5.0 and now long-time kernel hacker Con Kolivas is out with his 5.0-ck1 kernel patch as well as an updated MuQSS scheduler...

http://www.phoronix.com/scan.php?page=news_item&px=Linux-5.0-ck1-Released

[sh4nks]

I'd really be interested in a benchmark of this scheduler against the default CFS scheduler.

[yoshi314]

Honestly, i haven't used this scheduler since experiments with BFS. Is it much better compared to that one?

[r08z]

Honestly, i haven't used this scheduler since experiments with BFS. Is it much better compared to that one?

It's the almost the same but with improvements. This aims to be tickless and runs at 100HZ compare to BFS's 1000HZ but with guarantee latency deadlines. It's way better than the kernel's default CFS on heavy loads and interactive multitasking operations. The only problems is that older more reliable kernels are not supported. So you'll need to run this scheduler on the latest stable release.

[Axl_Mas]

Is it possible to have a review and benchmarks between CFS, MuQSS and PDS ( https://cchalpha.blogspot.com/ )?
From it's fork from MuQSS, PDS scheduler has been under hard developement and it seems very interesting!

[Cybmax]

Is it possible to have a review and benchmarks between CFS, MuQSS and PDS ( https://cchalpha.blogspot.com/ )?
From it's fork from MuQSS, PDS scheduler has been under hard developement and it seems very interesting!

Not huge differences between PDS and MuQSS, but the whole -ck patch series i guess give a wee bit more to it. So i would recommend doing the whole -ck patchset if you are comparing tho.
However Alfred Chen has launched a new scheduler - BMQ - BitMap Queue. Seems interesting tho, and supposedly a wee bit "lighter" than PDS. Will be interesting to follow the development of that tho.

[hax0r]

Con as always is the man, would we awesome if he stepped on and agreed to maintain MuQSS in the Linux kernel, then we would see its inclusion in Linux kernel upstream, would be a big win for desktop users. MuQSS and F2FS gives me great desktop experience and low latency, while CFS with BTRFS performs the worst (audio lags & mouse stutters under heavy disk I/O load). Stock Linux CFS scheduler is still optimized best for >1024 cpu-core workloads and desktop users are left out and have no easy choice.

[ArchLinux]

Con as always is the man, would we awesome if he stepped on and agreed to maintain MuQSS in the Linux kernel, then we would see its inclusion in Linux kernel upstream, would be a big win for desktop users. MuQSS and F2FS gives me great desktop experience and low latency, while CFS with BTRFS performs the worst (audio lags & mouse stutters under heavy disk I/O load). Stock Linux CFS scheduler is still optimized best for >1024 cpu-core workloads and desktop users are left out and have no easy choice.

Bro, you realize the reason BFS/MuQSS never made it mainline is because we're only every gonna have 1 CPU scheduler.

Or you guys will, I switched off of desktop Linux after about 10 years on and off. Leaves my name bit misleading.