Making The AMD Ryzen Threadripper 3990X Run Even Faster - By Loading Up Intel's Clear Linux

[CochainComplex]

I didn't dig into creating kind of an overlay repository for ffmpeg and friends, yet, like Debian-multimedia. I don't even know if swupd would support that.

While flatpak is great, it renders using clearlinux a little useless, as the apps come with their own libs, which don't have clearlinux's magic sauce.

sure magic sauce is gone but at least you have a workaround and mostly this are the applications which can be classified as desktop/multimedia. So the impact might be not so extreme.

and there is also the rpm/dnf workaround which is used for chrome, brave... the application it self has not the magic sauce but the dependencies..still a win

[oleid]

Maybe one should apply those optimizations to the flatpak runtimes?

[oleid]

I didn't dig into creating kind of an overlay repository for ffmpeg and friends, yet, like Debian-multimedia. I don't even know if swupd would support that.

While flatpak is great, it renders using clearlinux a little useless, as the apps come with their own libs, which don't have clearlinux's magic sauce.

[CochainComplex]

Clearlinux is really great on my desktop. The only mood point is not having a proper ffmpeg out of the box.

indeed, for sure not a beginner's distro. luckily it supports flatpak so missing applications can be pulled there without having to compile everything by hand.

their autospec toolchain is also interesting especially if there is a package existing and you just want to use a newer version. might be highlighted by swupd verify and overwritten if repair has been applied. but this is not as severe as it might sound.

[Buntolo]

> Coming most often in last place was Manjaro Linux and Fedora Workstation 31.

Hmm, I'm less inclined to jump on the Arch bandwagon now.

1) The gap between first and last is very thin

Clear Linux was 14% faster than the slowest distribution

2) Archlinux is not directly performance-oriented (not out of the box) but rather simplicity-oriented: you make (as in "assembly", not as in "compile") your system.
3) Clear Linux may not be the right choice for your PC, as a lot of stuff have been stripped and this may render it wimp on certain desktop platforms (missing support for not-so-older CPUs, libraries, etc.). Also I don't know about its support and community (check the archwiki or how much support there's for CentOS or Ubuntu)
4) Regardless of the previous entries: I suggest to just stick with what you already have. Contemplate a distro-swap only if you've free time to spend, you're unsatisfied with what you currently have, your current distro performs really bad/have other issues, etc. Basically if now everything's fine, stay like this.

Other than that, the System76 Thelio cases are very beautiful, do they sell them standalone?

[CochainComplex]

Because Intel doesn't release patches for most of their optimizations and many of them are even hardcoded. Almost all of them would have to be manually duplicated.

have a look at their pkg repos you can find the patches and just apply them as you wish on other distros. I did that in the past multiple times with openblas.

the reason why a lot of them are not mainlined is rather simple...they will restrict code funcionality to most recent cpus. So old <=bulldozers, <westmere will not be usable if there is not explicitly a cpu dispatcher handling this cases.

[oleid]

Clearlinux is really great on my desktop. The only mood point is not having a proper ffmpeg out of the box.

[duby229]

Newbie question regarding this topic, but if Clear Linux kernel optimizations run so well for both Intel and AMD, why isn't the default in other x86_64 distros?

Because Intel doesn't release patches for most of their optimizations and many of them are even hardcoded. Almost all of them would have to be manually duplicated.

[Ifr2]

Newbie question regarding this topic, but if Clear Linux kernel optimizations run so well for both Intel and AMD, why isn't the default in other x86_64 distros?

[nuetzel]

It is interesting how the brain interprets input according to our expectations.

Consider again the sentence I quoted in context of the article.

...across all the CPUs tested.

Sure. => I've to go back and have to learn _read_.