XanMod's Linux 5.10 Kernel Helping Tap Extra Performance With The AMD Ryzen 9 5900X

[pal666]

you can get any ordering by careful selection of benchmarks

[ms178]

Just from my own experiments, the right Kernel and Kernel configuration can have a huge impact on gaming, so to anyone who is willing to experiment, do it. I still remember the time where the out-of-the-box experience was significantly worse than my custom config (resulting in a 48 fps vs. 69 fps difference). I suspect the default governor was the culprit as that particular older game is very CPU sensitive and changing it to the "performance" governor was one of my configuration adjustments.

[ext73]

It can be faster than xanamod - especially in games - native and Windows > Premium kernels > netext73.pl

[Vistaus]

Yes, you can use udev to force your nvme drive to use "none" as your IO scheduler. However, if you're doing things that need that throughput, perhaps you should not be using Liquorix (and Zen Kernel by extension)? Liquorix is configured for responsiveness and low latency behavior at the cost of power consumption (and some throughput). If it wins any throughput benchmarks it's almost by accident . It's for systems that are multitasking and the multitasking behavior must produce the least amount of side effects on neighboring processes.

If you're at the point where you're tuning options to favor throughput, something you'd expect for systems treated like servers/appliances, I'd go with XanMod. It makes enough significant changes that it's a no brainer if you want a free upgrade in those aspects. Just know that throughput is not what makes a system feels fast, it's responsiveness. That's why you don't typically run low latency kernels on servers since there's no way to "feel" the difference, so just go all in on throughput.

And one final note on the git benchmark. Most git operations are single or low threaded operations. MuQSS really doesn't do a good job keeping those threads on the same cores for fairness reasons, and in turn, cpufreq doesn't know how to set the right frequency. The only solution I've found to that is to just tell cpufreq to go full speed at 45% core utilization. It works most of the time but it's best to run the performance governor otherwise. Most modern CPUs idle very efficiently so you can get away with that. For laptops though, stick to ondemand.

EDIT: Just noticed that this time around, Michael stuck with the default kernel configuration. Liquorix in these test was using the performance governor. Seems like in this case, MuQSS just can't keep git on the same core to save its life and loses on throughput there.

ondemand on laptops? No way. ondemand doesn't do that great of a job on laptops. schedutil is a much better and modern replacement.

[Vistaus]

It can be faster than xanamod - especially in games - native and Windows > Premium kernels > netext73.pl

What exactly do you mean?

[damentz]

ondemand on laptops? No way. ondemand doesn't do that great of a job on laptops. schedutil is a much better and modern replacement.

I'd like schedutil to work well, but on MuQSS it behaves exactly like performance. If you're on Intel you can force enable intel_pstate on Liquorix to get power savings, but your desktop will feel noticeably slower since it wants to pick lower frequencies even when there's work to be done.

[ext73]

What exactly do you mean?

e.g. ... on new kernels will be Ultra in 24-35 fps:

https://youtu.be/X-CSAcitHfs

[Linuxxx]

e.g. ... on new kernels will be Ultra in 24-35 fps:

https://youtu.be/X-CSAcitHfs

There are multiple things wrong with your self-proclaimed "premium" kernels:

- Your disabling security mitigations by default!
- Your compiling them with unfinished/alpha-level compilers with known regressions! (GCC 11 in this case)
- Your kernel isn't even configured to be fully-preemptible! (PREEMPT missing when executing »uname -a«)

Also, if your kernels really could magically run Windows games faster on Linux than on Windows itself, maybe you should consider applying to CD Projekt RED & help out your fellow Polish-countrymen?
I bet they really are desperately in need of a guy like you at the moment...

[ext73]

There are multiple things wrong with your self-proclaimed "premium" kernels:

- Your disabling security mitigations by default!
- Your compiling them with unfinished/alpha-level compilers with known regressions! (GCC 11 in this case)
- Your kernel isn't even configured to be fully-preemptible! (PREEMPT missing when executing »uname -a«)

Also, if your kernels really could magically run Windows games faster on Linux than on Windows itself, maybe you should consider applying to CD Projekt RED & help out your fellow Polish-countrymen?
I bet they really are desperately in need of a guy like you at the moment...

Hi

ad. 1 yes, I turn the mitigation on and off - depending on the needs - with two clicks

https://www.dropbox.com/s/llmnyj9jl7...93941.jpg?dl=0

ad.2 Yes, I am compiling a kernel using Gcc 11 and despite the regressions still in existence, the kernels are very stable and I am able to get code performance unattainable for Gcc 10.2.x or LLVM/Clang 11

ad.3 This 'argument' has already pushed me into the ground hehe - compare gaming and reaction speed, e.g. running the application on my kernels vs Full PREEMPT - I built and shared PREEMPT kernels about 10 years ago. Currently, such a solution is pointless - except for RT kernels. In addition, the clock used in the Xana is 500 Hz ... I tested about 7 years ago ... now I go in the opposite direction and the latest kernels will probably have a clock of 125 Hz

https://www.dropbox.com/s/brsh4eaks4...95726.jpg?dl=0

btw. as for REDs, yes they xxxxx ... but I think in 4-6 months they will make C2077 a game worthy of expectations. Here, unfortunately, the pressure from investors turned out to be stronger than common sense.

And as for the C2077 performance - under the latest test kernel, I manage to achieve it on the indicated hardware - as it seen in the video - 24-38 fps in FullHD > Ultra

[pkese]

I have a new Ryzen 5900x and Ubuntu 2010's default 5.8 kernel.
With a couple of problems:
- the 2.5G Ethernet driver is missing in kernel
- sensors don't report any CPU temperatures
- EDAC drivers are missing, etc.

Is this to be expected? These CPUs have been introduced a few months ago and these drivers are available (in later kernels).
Why wouldn't Ubuntu bother to backport these drivers to their kernels?