2017: Should Linux Benchmarking Still Be Mostly Done With Ubuntu?

[Michael]

It would be interesting to see a community based team that would test games/compilers etc etc on their own rigs to submit to Phoronix. That way we get a bit of a mix using real world game/production rigs.

Anyone already can and do using OpenBenchmarking.org.

[Hibbelharry]

Anyone already can and do using OpenBenchmarking.org.

That could really work, if you would orchestrate things a little. Why don't you try something like:

Community please test Game xyz with profile abc in resolution FGH. You would then be able to aggregate the results and offer commented "swarm benchmarks results".

[duby229]

That could really work, if you would orchestrate things a little. Why don't you try something like:

Community please test Game xyz with profile abc in resolution FGH. You would then be able to aggregate the results and offer commented "swarm benchmarks results".

I think that would end up as a fruit salad if you know what I'm sayin. But maybe fruit salad is a good thing.

[dungeon]

Since didn't work for arch as it contains no branding at all in terms of the user agent.

Does not change a point much really, even if we pretend to be very optimistic and count all 4% of unknown as Arch

[cjcox]

Ubuntu is fine. Might be nice it were a combo. Ubuntu (for novices or the ignorant), OpenSuse Tumbleweed (view of what's to come), CentOS "latest" (yestetday's predominent player in the enterprise). Noting that both the latter two can be very different from Ubuntu and tend to reveal problems with OpenBenchmark and it's documentation. If just "gaming", you can skip the CentOS runs. Obviously Tumbleweed is hit or miss on the gaming front especially for proprietary drivers (but ok for latest open source drivers... without doing custom work against something like Ubuntu).

Just my two cents. Friends don't let friends run Ubuntu... but we're kind of used to seeing it. I can "translate" if it stays the "only one".

[Darknation]

Anyone already can and do using OpenBenchmarking.org.

Very true. But I was looking at it in a more productive way. People you would bring on to do some of the testing for you outside of what resources might be limited too. I'm probably not wording this very well as I'm half asleep. But it might take some of the burden off your shoulders and give the community more incentive to help be more proactive. Not that any standing members here are the otherwise.

[Dukenukemx]

It was same as Firefox, anyway Iceweasel doesn't exist anymore for about year
RIP Iceweasel, 13 Nov 2006 – 10 Mar 2016

Shows how long I've used Debian. Even on my RPI3 I use Ubuntu Mate.

Debian won't add support for Ubuntu PPAs, because Launchpad need to add support for Debian first - it does not work the opposite way . Things need to be builded in proper Debian enviroment to qualify to be added and to can be recommended...

You know what I mean. Add support for their Launchpad PPA system. It's a really nice system to have.

meanwhile no one sane advise you to use Ubuntu's PPAs on Debian as is

I'm not a very sane person.

[dungeon]

You know what I mean. Add support for their Launchpad PPA system. It's a really nice system to have.

I know what you want but It is impossible to add that support, because that is Ubuntu's PPA supporting Ubuntu base only and not Debian base... not sure why you don't understand that.

I'm not a very sane person.

Ah, i am starting to guess why is that

[cj.wijtmans]

Its not just compilation, keeping up to date is a bitch too.
Sources: personal experience.

Here i was thinking that updating on gentoo is extremely slow. GCC6 is not even in unstable branch yet (have to unmask the package). Left my computer for a few weeks and only a few software updates.
Sources: personal experience(managing both my gentoo desktop and several gentoo servers)

[philnc]

I'm late to the party as usual... I'd say testing the latest Ubuntu AND Fedora would provide the best overall coverage for most. My particular interest is servers, but my primary personal desktop right now is Fedora. But in the past I've often recommended Ubuntu to others. Depending on what you're measuring, results can diverge: like Ubuntu using x.org and Fedora with Wayland. Of course if the particular benchmark wouldn't show a significant difference, I can't see any problem with publishing only the Ubuntu results.