sched-ext Tutorial

Tento obsah zatím není dostupný ve vašem jazyce.

Extensible Scheduler Class, or better known as sched-ext is a Linux kernel feature which enables implementing kernel thread schedulers in BPF (Berkeley Package Filter) and dynamically loading them. Essentially this allows end-users to change their schedulers in userspace without the need to build another kernel for a different scheduler.

Starting and using the scx schedulers

You can find the schedulers in the scx-scheds or scx-scheds-git package. Simply run following command to install the package:

sudo pacman -Sy scx-scheds

Starting the Scheduler

The scheduler can be simply started in the terminal with following command:

sudo scx_rusty

This will launch the rusty scheduler and detach the default scheduler.

To stop the scheduler, you simply run CTRL + C and the scheduler will be stopped and the default kernel scheduler will be used again.

Systemd Service

The scx packages provides also a systemd service. This service can be configured in /etc/default/scx. You can change the scheduler used by sched-ext and set custom flags for each scheduler in this configuration file. By default, the services uses the rusty scheduler. If you want to change the scheduler used by the service simply change the SCX_SCHEDULER=scx_rusty to SCX_SCHEDULER=scx_lavd or others.

Now you can start/enable/stop the scheduler as any other systemd service.

Use the scx scheduler as default and enable directly at the boot

sudo systemctl enable --now scx

Start the scheduler only once via the systemd service

sudo systemctl start scx

Stop the scx scheduler via systemd service

sudo systemctl stop scx

For more information: Sched-ext systemd service

CachyOS Kernel Manager

The scx schedulers can be accessed and configured through the GUI.

Introduction to the main schedulers

Since there are many schedulers to choose from, we want to give a little introduction about the schedulers in hand.

Feel free to report any issue or feedback to their GitHub referenced below.

scx_bpfland

Developed by: Andrea Righi (arighi GitHub)

A vruntime-based sched_ext scheduler that prioritizes interactive workloads. Highly flexible and easy to adapt.

Bpfland when making decisions on which cores to use, it takes in consideration their cache layout and which cores share the same L2/L3 cache leading to fewer cache misses = more performance.

Use cases:

Gaming
Desktop usage
Multimedia/Audio production
Great interactivity under intensive workloads
Power saving

scx_flash

Developed by: Andrea Righi (arighi GitHub)

A scheduler that focuses on ensuring fairness among tasks and performance predictability. This scheduler is introduced as a replacement of the “lowlatency” mode in scx_bpfland.

Use cases:

Gaming
Latency sensitive workloads such as multimedia or real-time audio processing
Need for responsiveness under over-stressed situations
Consistency in performance

scx_lavd

Developed by: Changwoo Min (multics69 GitHub).

Brief introduction to LAVD from Changwoo:

LAVD is a new scheduling algorithm which is still under development. It is motivated by gaming workloads, which are latency-critical and communication-heavy. It aims to minimize latency spikes while maintaining overall good throughput and fair use of CPU time among tasks.

Use cases:

Gaming
Audio Production
Latency sensitive workloads
Desktop usage
Great interactivity under intensive workloads
Power saving

One of the main and awesome capabilities that LAVD includes is Core Compaction. which without going into technical details: When CPU usage < 50%, Currently active cores will run for longer and at a higher frequency. Meanwhile Idle Cores will stay in C-State (Sleep) for a much longer duration achieving less overall power usage.

scx_rusty

Developed by: David Vernet (Byte-Lab GitHub)

Being one of the heaviest schedulers yet released on sched-ext, it comes with a lot of features that add to his flexibility and capability. Tunability is one of them so you can adjust Rusty to your desires and use case.

Use cases:

Gaming
Latency sensitive workloads
Desktop usage
Multimedia/Audio production
Latency sensitive workloads
Great interactivity under intensive workloads
Power saving

For more information about what can be done with Rusty and his tunable flags. Check out the help page:

scx_rusty --help

General recommendations

LAVD Autopilot & Autopower

Quotes from Changwoo Min: In the autopilot mode, the scheduler dynamically changes its power mode (Powersave, Balanced or Performance) according to system’s load (CPU utilization).

Autopower: Automatically decide the scheduler’s power mode based on the system’s energy profile aka EPP (Energy Performance Preference).

# Autopower can be activated by the following flag:
--autopower
# e.g:
scx_lavd --autopower

Disable ananicy-cpp

When using any of the schedulers from the sched-ext framework, it’s strongly advised to disable and avoid ananicy-cpp due to causing conflicts because it amplifies the priority gap that the scheduler is already creating/handling by boosting interactive tasks creating an excessive prioritization starving other tasks increasing the chance of hitting the sched_ext watchdog timeout kicking the running scheduler aka causing an stall.

In order to disable/stop ananicy-cpp, run the following command:

systemctl disable --now ananicy-cpp

FAQ

Why X scheduler performs worse than the other?

They’re lots of variables that take place when comparing each one of them, for example: How do they measure a task’s weight? Does it prioritize interactive instead of non interactive ones? and so on.

Why everyone keeps saying this X scheduler is the best for X case but it does not perform as well for me?

Similar to the answer from above. Which cpu is used and his design, being their core layout or similar might cause the scheduler to not work as intended.

That’s why having choices is one of the highlights from the sched-ext framework, so don’t be scared to try the main ones and see which one works best for your use case, being ex: fps stability, maximum performance, responsiveness under intensive workloads etc.

Each of these schedulers’ behavior can be tuned with flags. Refer to each scheduler’s --help output for a brief explanation of what each flag does

# Example:
❯ scx_lavd --help

Options:
      --no-core-compaction
          Disable core compaction, which uses minimum CPUs for power saving, and always use all the online
          CPUs

      --prefer-smt-core
          Use SMT logical cores before using other physical cores in core compaction

      --no-freq-scaling
          Disable frequency scaling by scx_lavd

The use cases of these schedulers are quite similar… why is that?

Mainly because they are (for now) multipurpose schedulers, meaning they adapt to many workloads even if they don’t excel at all of them.

In order to find out which one fits you best, there is no other shortcut than to test it yourself.

I’m missing a scheduler that some users are mentioning or testing in the Discord server

The reason behind is probably because these schedulers are being tested therefore they’re only included as a cherry pick in the scx-scheds-git package instead of the stable version.

Learn More

If you want to learn more about the sched-ext framework. Take a look at the links below.