Kernel liquorix - any real difference ?  [Solved]

Message
Author
User avatar
~FLOW~
Posts: 294
Joined: Thu May 19, 2022 12:43 pm

Kernel liquorix - any real difference ?

#1 Post by ~FLOW~ »

Greetings my excellent friends,
I notice that now we have the liquorix option in the kernel list, to install.
Did anyone notice real difference in performance with this one ?
I coudn't find any thread here that is discussing about this.
MX-23.4 / KDE Plasma 5.27
Kernel: 6.XX

Be patient !!
What seams easy for you now, for a newbie might be a monumental helping hand.
An efficient pack will move as fast as the slowest member.

User avatar
wdscharff
Posts: 1162
Joined: Mon Feb 24, 2020 1:07 am

Re: Kernel liquorix - any real difference ?

#2 Post by wdscharff »

There were a few discussions with measurements in the FB group last year.
It depends on your applications, it's as simple as that.My applications are image & video editing and my daily backup via snapshot.
Biggest difference between the alternative kernels and the "normal" one is the lower resource consumption, with Fluxbox a difference between <300mb with xanmod/liquorix to >400mb with mainstream kernel. Video rendering and compression (zstd and xz) that scale well to multi-processor have a speed gain of 1-3%, sometimes xanmod is ahead, sometimes liquorix. Is it worth it?
Yes, because it's pretty low-risk. a new kernel is installed quickly (at least on my 12-core) or can be removed again with mx, even if the 1-2% speed advantage isn't really noticeable in some daily applications, but it doesn't cost anything, I've been using xanmod and liquorix kernels without exception since mx19, the MX kernel is only on my system as a fallback (never needed).

you can also easily answer such questions yourself by trying them out, which has the advantage that the answer fits your own hardware and your own application programs, because there are differences (on my Intel laptop, the speed differences are rather at the measurement limit, there is only the advantage of the lower ram requirement).
my working horse Desktop AMD Ryzen 9 3900x, 32GB Ram // SSD ... enough
mx-fluxbox, what else?

In nature there are neither rewards nor punishments.
There are consequences.


my wallpaper gallery

User avatar
baldyeti
Posts: 732
Joined: Sat Dec 05, 2009 3:37 pm

Re: Kernel liquorix - any real difference ?

#3 Post by baldyeti »

Another difference is regarding security updates - only the standard debian kernel receives them. The liquorix/ahs/antiX ... kernels do not. KDE is normally an AHS release but the initial MX23 ISO ships withe debian 6.10 kernel; just install the "linux-image-amd64" package to ensure you will be getting future security updates.

MXRobo
Posts: 1843
Joined: Thu Nov 14, 2019 12:09 pm

Re: Kernel liquorix - any real difference ?

#4 Post by MXRobo »

Liquorix is an enthusiast Linux kernel designed for uncompromised responsiveness in interactive systems, enabling low latency compute in A/V production, and reduced frame time deviations in games. per: https://liquorix.net/
A review of liquorix, and XanMod of which I'm unfamiliar. https://www.phoronix.com/review/ryzen5-xanmod-liquorix

User avatar
CharlesV
Administrator
Posts: 8180
Joined: Sun Jul 07, 2019 5:11 pm

Re: Kernel liquorix - any real difference ?

#5 Post by CharlesV »

My general observation is that the liquorix kernels seem to fix some issues with various things: sound, mice, some keyboards, and some visual issues as well.

My personal observation is that ALL of my test and production machines runs faster, "smoother" and with less issues. Even the older models. I have seen specifically the 6.2 kernel fix booting and stability issues on some machines, mouse, touchpad and even resolve some video issues.

Something I have not seen anyone mention above, the liquorix kernels seem to run faster. (I was told it is because there is no throttling in the kernel, but I cannot confirm this.) However, there is no question that it runs faster in my testing.
*QSI = Quick System Info from menu (Copy for Forum)
*MXPI = MX Package Installer
*Please check the solved checkbox on the post that solved it.
*Linux -This is the way!

User avatar
~FLOW~
Posts: 294
Joined: Thu May 19, 2022 12:43 pm

Re: Kernel liquorix - any real difference ?

#6 Post by ~FLOW~ »

Thank you for your input, CharlesV
MX-23.4 / KDE Plasma 5.27
Kernel: 6.XX

Be patient !!
What seams easy for you now, for a newbie might be a monumental helping hand.
An efficient pack will move as fast as the slowest member.

User avatar
Stevo
Developer
Posts: 14922
Joined: Fri Dec 15, 2006 7:07 pm

Re: Kernel liquorix - any real difference ?

#7 Post by Stevo »

The standard Debian kernel is tuned more for server use than what a single user might desire.

This makes Liquorix more suitable for situations where a lower latency is required, such as multimedia creation. That's why AV Linux ships with a Liquorix kernel. Gamers may get faster responses, too.

They also enable more hardware driver builds inside the kernel than the more conservative Debian approach.
Last edited by Stevo on Sun Aug 20, 2023 12:39 pm, edited 1 time in total.
MXPI = MX Package Installer
QSI = Quick System Info from menu
The MX Test repository is mostly backports; not the same as Debian testing

User avatar
~FLOW~
Posts: 294
Joined: Thu May 19, 2022 12:43 pm

Re: Kernel liquorix - any real difference ?

#8 Post by ~FLOW~ »

Stevo wrote: Sun Aug 20, 2023 11:41 am The standard Debian kernel is tuned more for server use
uuuuuu, good point, thank you
MX-23.4 / KDE Plasma 5.27
Kernel: 6.XX

Be patient !!
What seams easy for you now, for a newbie might be a monumental helping hand.
An efficient pack will move as fast as the slowest member.

User avatar
~FLOW~
Posts: 294
Joined: Thu May 19, 2022 12:43 pm

Re: Kernel liquorix - any real difference ?  [Solved]

#9 Post by ~FLOW~ »

Still,
on a low budget laptop I didn't feel any difference. So it's no for me.
MX-23.4 / KDE Plasma 5.27
Kernel: 6.XX

Be patient !!
What seams easy for you now, for a newbie might be a monumental helping hand.
An efficient pack will move as fast as the slowest member.

User avatar
m_pav
Developer
Posts: 1916
Joined: Sun Aug 06, 2006 3:02 pm

Re: Kernel liquorix - any real difference ?

#10 Post by m_pav »

Example from a HP 8300 SFF G2 Desktop running a 6th Gen Intel pushing 2 x 1080P displays in extended Desktop mode. Modesetting driver with standard kernel and no tweaks whatsoever = significant pauses when streaming 1080p Live TV to an external projector through HDMI. Using MX-Tweak, changed the driver to Intel, much better, but still noticeable screen tearing.
Changed to Liquorix kernel, no faults whatsoever, clean fast video through HDMI.

That got me thinking, we have a rudimentary benchmarking tool, System Profiler and Benchmark. I tested the last item in the list of benchmarks it provides, GPU drawing, to see what difference each option made. I know it's rudimentary and not particularly relevant to real world results, but it still provides actual numbers vs an opinion.

Standard kernel, modesetting driver, score was roughly 4600
Standard kernel, Intel driver, the score was roughly 6500
Liquorix kernel, Intel Driver, the score was about 12600

So we see certain elements of the system are improved, but where there is a gain in one area, there is a loss in another because a machine can only do as much as it can handle.

The Stock Debian kernels run at (I think) 250hz which is well optimised for Servers. This means the kernel will spend more time on each instruction before it is interrupted to check for other queued instructions. This results in better performance in throughput so for example, server side video encoding will be much faster, it is processor, memory and storage intensive.

The Liquorix kernels run at 1000hz, so the system feels more responsive on the Desktop, but it has less available time to work on each instruction before it is interrupted to check for other queued instructions. When a single task is in operation, such as the graphics benchmark test above, the score appears to show the system as being more responsive, and it is for this process, but the overall throughput will take a hit because the system is stopping 3x more per second, therefore usage case scenarios are an important consideration.

It all depends on what you're using your system for. If you game or you need to do real-time audio/video, then 1000hz may be a better option, but Liquorix kernels go much farther than just the kernel tick. In my scenario, my Lenovo T560 is a CTO build, so it doesn't conform to their run of the mill builds and as such, it has some more exotic options. This makes it more susceptible to faults with sleep states and using a Liquorix kernel on this machine sorts out enough of the quirks to make it quite pleasurable to use, but I notice a drop in performance when copying large volumes of data. The trade-off is worth it.
Mike P

Regd Linux User #472293
(Daily) Lenovo T560, i7-6600U, 16GB, 2.0TB SSD, MX_ahs
(ManCave) AMD Ryzen 5 5600G, 32G, 8TB mixed, MX_ahs
(Spare)2017 Macbook Air 7,2, 8GB, 256GB SSD, MX_ahs

Post Reply

Return to “General”