Fedora vs. Ubuntu vs. openSUSE vs. Clear Linux For Intel Steam Gaming Performance

[eydee]

Getting there, still, a normal desktop version, intended for simple end-users for daily use is needed. In this state it still isn't worth messing with for playing games.

[indepe]

Getting there, still, a normal desktop version, intended for simple end-users for daily use is needed. In this state it still isn't worth messing with for playing games.

Fedora (recompiled) added to Clear Linux...

[MadCatX]

@Michael: Here is an idea for a benchmark: Fedora 25 running on bare X instead of Wayland, Ubuntu 16.10 with latest Mesa and CPUFreq governor against Clear Linux.

[rabcor]

Clearlinux is impressive.

[dungeon]

@Michael: Here is an idea for a benchmark: Fedora 25 running on bare X instead of Wayland, Ubuntu 16.10 with latest Mesa and CPUFreq governor against Clear Linux.

Yup power managment totally disabled and xinit of all Otherwise i am not sure we benchmarking games here, but compositors and power managment differences

Overclockers and people who wanna every bit of performance even disable font antialiasing, that is how far that goes disable every uneeded process, deamons, disable even tick to reduce possible stutter (including local time clock and any kind of uneeded measurements... Playstation 4 is not PC because of perf purpose ), etc... more you disable better for you Should I say... they also don't use openbenchmarking, as that probably bite some resources for sure

[mlau]

It would be interesting to bench against a similarly up-to-date distro like Arch Linux.

Nah, Gentoo, built with all ricer options enabled for the host cpu. It'll probably be as fast as Clear Linux.

[Vistaus]

Thank you for correcting a false assumption

Please keep up the good work and hopefully other distros can pick up some of the work you are doing

It's already being picked up: Solus (Linux distro) is using some of Clear Linux's optimizations.

[duby229]

Nah, Gentoo, built with all ricer options enabled for the host cpu. It'll probably be as fast as Clear Linux.

Not really. I've literally tried ricing Gentoo hard. The biggest improvement is LTO, but it's way too buggy yet. Otherwise ricing Gentoo doesn't really help. Maybe just slightly, but the only thing you'll likely notice after ricing is new runtime bugs.

The only thing I do nowadays is -o2 -pipe -march=native -mtune=native. It'll be nearly 100% successful and achieve at least 98% the performance of a highly riced setup.

[mlau]

Not really. I've literally tried ricing Gentoo hard. The biggest improvement is LTO, but it's way too buggy yet. Otherwise ricing Gentoo doesn't really help. Maybe just slightly, but the only thing you'll likely notice after ricing is new runtime bugs.

The only thing I do nowadays is -o2 -pipe -march=native -mtune=native. It'll be nearly 100% successful and achieve at least 98% the performance of a highly riced setup.

I use O3 for almost all, it's safe enough nowadays, lto for a few select packages that actually work with it and give a subjective performance gain (mesa is one of those few).
For fun I built a git head snapshot of mesa for mips, and lto improved performance of software-rendered glxgears from 9fps to 11fps. Not much, but still impressive
considering the 700MHz mips has no fpu and no gfx accelerator.

[leipero]

Maybe a bit off topic, but I just noticed something and I'm afraid i will forget to mention when apropriate, so my apologies. I love tor ead reviews/comparison of GPUs and drivers, but when testing steam games, it's not noted in any (?) article if you are using steam with native libraries or steam runtime. I think it would be useful for readers to include such information, and maybe even do a comparison if there's any performance diference between the two?