AlmaLinux, CentOS Stream, Clear Linux, Debian, Fedora & Ubuntu On AMD 4th Gen EPYC Genoa

[andyprough]

I really don't understand why do some people claim Fedora is the slowest distro. As seen in this and earlier benchmarks that I remember, it's average+ compared to all other distros.

Because it used to suck. Now it just doesn't suck as bad as Ubuntu. #LifeAchievements.

[Danielsan]

Debian kind of sucks and because of that all Debian and Ubuntu based distros too.
Too bad that Debian 12 that will be supported for so many years still doesn't come with some good performance improvements.

I don't see Debian sucking by any means, nobody is optimizing or gearing Debian for any specific task, however it always performs better than Ubuntu.

Also this comparison is misleading, all those distros but Debian are baked by a sponsor, and all of them but Debian are meant to be enterprise distributions for server and cloud task and are optimized for such purposes.

[pmorph]

Unfortunately Clear Linux is not really usable as a regular desktop OS for the average user. I do think we need at least one distro that is performance oriented, but at the same time adequate as a desktop OS. AFAIK, this is what Serpent OS will try to do.

Performance oriented server can be rather a different thing than a performance oriented desktop. Chances are it will suck at some use case anyway.

[Britoid]

Rocky and Alma should be bug-to-bug clone with RHEL, thus the results should be practically the same.

CentOS Stream (~9.2) and Alma (9.1) show very similar results. Oracle with their updated kernel would be interesting though.

What surprises me is how Fedora lags in these tests compared to Enterprise distros.

Getting anything changed in Fedora requires 10 levels of beurecracy.

[sarmad]

Can someone remind us why Clear Linux always beats the competition with a big margin and why after all those years no other distro decided to apply the same optimizations that Clear Linux does?

[Calinou]

Can someone remind us why Clear Linux always beats the competition with a big margin and why after all those years no other distro decided to apply the same optimizations that Clear Linux does?

Defaulting to the "performance" governor on desktop PCs is a big part of why Clear Linux performs better out of the box. The downside is higher power consumption, especially at idle (which is why it's not done on laptops by default). You can see how close the Ubuntu "perf" result is to Clear Linux on the 7950X benchmark: https://www.phoronix.com/review/zen4-clear-linux

You can do this on other distributions too (it's a single command to run on login: "sudo cpupower frequency-set -g performance").

On Windows, a similar thing can be done by opting into the High Performance power profile.

[sarmad]

Defaulting to the "performance" governor on desktop PCs is a big part of why Clear Linux performs better out of the box. The downside is higher power consumption, especially at idle (which is why it's not done on laptops by default). You can see how close the Ubuntu "perf" result is to Clear Linux on the 7950X benchmark: https://www.phoronix.com/review/zen4-clear-linux

You can do this on other distributions too (it's a single command to run on login: "sudo cpupower frequency-set -g performance").

On Windows, a similar thing can be done by opting into the High Performance power profile.

Do all other distros in this benchmark default to powersave governor? There is a big difference between Clear and the second in this list.

[elatllat]

The env img Needs more bleed/margin to not have the writing cut off as fonts in svg files render diffrent on every OS... or txt to curve it.

[kylew77]

These are some cool benchmarks, but may I ask what ever happened to the Raptor Lake OpenBSD benchmarks?

[NotMine999]

In Michael's house they don't throw another log on the holiday fire.

They fire up another server and start another batch of Linux tests.