Ubuntu 20.04 LTS vs. Clear Linux On The Intel Core i9 9900KS, AMD Ryzen 9 3900X

[lsatenstein]

I am a desktop user. Half my retired day is browsing the web, doing some test compiles of software, testing other software and watching audio-visual presentations. For youtube videos, CL is on a par with other distros, but when friends send me mp4's or *.mov's or other audio-videos, CL offers no option. I am asked to compile my own VLC version. My attitude is "why bother", I have Fedora 32 on a second disk, and it is complete for audio-visual support, So CL, until I want to update my C code, its casual visits or visits to compile programs that execute faster and are smaller. Programs are ones I compile for my own uses.

[deusexmachina]

Love your RTFM comment. All GUI's are bad - just RTFM!
Performance is only good, if it can perform RTFM is not performance. RTFM is not performing. Ergonomic design prevents errors. Prevention is better than cure.

As you can see others agree and like my comment. You turned it into something inflammatory, but it is not. Clear Linux is a distro with pre-setup kernel optimizations that can be replicated in other distros - probably exceeded. Your comment barely makes sense and you have the attitude that you shouldn't even have to explain yourself.

[gregzeng]

(1) At first we might expect the Intel CPU to be a clear winner for the Intel-designed operating system. Not so. Seems that AMD matches the Intel CPU, generally.
(2) The other surprise is that the Clear Linux is already speed optimised to better perform, but only on bench tests. If the Ubuntu operating system is similarly optimised (latest low-latency kernel, released officially by Ubuntu), the results could be quite different.
(3) There are several other tweaks that can optimise computer operating systems. These tweaks however compromise other aspects of the operating system, such as when bench testing is not being done, which is most of the time.
(4) Users of operating systems know that more is needed than remote, academic bench tests. What surprises me is that the upgrades to the operating system are not mentioned.
(5) In the installation of good operating systems, such as Ubuntu, the administrator needs to give permission for the Linux kernel, etc to be upgraded. Depending on the attention, it is very rare that the Linux kernel used in the first installation will remain in the operating system soon after the system is used.
(6) Ubuntu 20.04 is the alpha-release of the forthcoming operating system. It is not final, so has some designed in "slow-downs" to assist debugging & trouble-solving. The other operating system AFAIK is a final release, designed for bench tests only, with slow-downs removed.

[gregzeng]

No one uses Ubuntu for performance... I don't understand why Clear Linux keeps getting put against lame defaults. Try Gentoo with their wiki's performance recommendations...

Love your RTFM comment. All GUI's are bad - just RTFM!
Performance is only good, if it can perform RTFM is not performance. RTFM is not performing. Ergonomic design prevents errors. Prevention is better than cure.

[Radtraveller]

Seems there are, at least, 3 levels of use cases that could be tested for each Linux environment.
Server, “power users” and users who just want something that works for their day to day desktop so they can ditch windows.

I have “lurked” phoronix for quite awhile and Michael does a great job. Just not sure whether he has the bandwidth to compare each and every Linux over different use cases.

Quite sure a self brewed gentoo or lfs, or any other lean, tightly integrated and optimized Linux would be faster than Ubuntu...

But, unfortunately (maybe) the regular user just wants something easy to install, no headache config of, especially video and WiFi systems, that will provide all the necessary daily use functionality without having to get, at least, the equivalent of a 2 year degree in Computer Science to manage compilations and command line optimizations..

i don’t use Ubuntu on a daily basis, but comparing the ease of install and the time it takes from booting the install to getting online and maybe playing a game vs clear or fedora and not having WiFi, especially on a laptop without a nic port for cabled connection....

Anyway, just a thought that ease of install with ease of hardware configuration is often a more important reason for people choosing a Linux flavor as a desktop than a few percent, or tenths of a percent, of performance advantage.

And the folks that want server flavor or the “power” users, already know how to choose and optimize their systems. (Often using sites like phoronix for data). Maybe download and run the phoronix test suite yourself and post the results for your gentoo install?
Others can check out the numbers themselves to prove or disprove what you post if they want.

[deusexmachina]

No one uses Ubuntu for performance... I don't understand why Clear Linux keeps getting put against lame defaults. Try Gentoo with their wiki's performance recommendations...

[nuetzel]

Now, retry with LVI.

[marciosr]

Since it's from Flatpak you have to run the program with --filesystem=/some/dir to access stuff outside of the sandbox. That's one of the biggest peeves I have with running Flats. I get why it's like it is, but with certain programs it isn't very helpful and can be a cause of issues and annoyances. There are a few other ones like --share and --socket that may have helped with the Krita issue.

There is a graphical app to help manage flatpak permissions: https://flathub.org/apps/details/com...chx84.Flatseal

[Radtraveller]

Yes but which one handles Optimus laptop graphics better?
Fedora still is a pain and difficult to configure.
I haven't tried Clear in awhile, but it didn't do well with my optimus system last time i attempted it.

[Michael]

It would be nice if there was a way to measure only the core / kernel performance, as it seems most of the advantages for clear come from the additional optimization of the external packages, rather than the os itself.

Maybe compile the tests with the same flags rather than using whatever the distro package manager offers out of the box.

And yes, I know that default packages is what most people would use, but there is certainly value in establishing the core performance for people who run proprietary software.

There are numerous tests there using the same exact binaries across platforms like like ArrayFire, Blender, V-Ray, IndigoBench, etc.