Vote On A Distribution For Linux Benchmarking

[GreatEmerald]

The number of people that the tests are relevant to is a factor though.

Agreed. But the number of people that the tests are relevant to are also the number of people voting in this very poll. Which means that the result of this poll is more reliable than any indirect sources you can find.

Okay but why is this relevant? Are you suggesting that Michael should use SUSEStudio to create a bootable CD or USB image for every benchmark?

It depends. If a pristine condition is required, then yes, it's much easier to achieve that using SUSEStudio. Swapping out a package is really easy. Just takes time to build and download, but allows anyone to verify it our test the package out. Which is also nice for direct hardware comparisons, since it cuts the software difference out of the equation. You could also prepare a pristine base: preinstall PTS, everything else that you need and you don't want to change, but don't install what you do want to change. Then when booted, install that component, then reboot, install the alternative, repeat. Makes it faster, and you always start fresh, there is nothing leftover from the previous run.

[synaptix]

VPN's and PGP keys it's the only way

Btw you Play DOTA 2?

Even VPN aren't 100% private.

[jrch2k8]

ok, for the next 15 pages of phoronix World FlameWar III.

add FreeBSD instead of ubuntu

[fedora winning yay]

[blackout23]

The level of butthurt on this thread from Ubuntu Fans is OVER 9000!!!!

[LinuxGamer]

ok, for the next 15 pages of phoronix World FlameWar III.

add FreeBSD instead of ubuntu

[fedora winning yay]

don't you know the next release of Ubuntu is going to be BSD

[jrch2k8]

don't you know the next release of Ubuntu is going to be BSD

would be nice if true, probably bsd model fit more canonical goals than Linux/GPL

[gerddie]

Debian testing

Instead of Debian sid, Debian testing would make a lot more: Unless it is in freeze (every two ore tree years or so) it follows sid closely but is more stable.

[peppercats]

Would have really liked to see Gentoo as it's the only one that's not completely similar to the rest, but alas, I guess I'll vote Fedora.

I'm not even sure why Ubuntu is an option though, it's essentially just Debian unstable with a couple extra packages(and unity).

[Bathroom Humor]

Obviously there is going to be a big argument over which OS is better for *subjective reasons*, but to be perfectly reasonable, I'd rather have the tests stick to what he normally uses, and maybe add Fedora just for due diligence.
Ubuntu isn't a huge mess. Not more than any other popular distro. Fedora and Ubuntu make the best choices for no other reason except they are the most popular distros who are fairly different from each other. Any linux distro will be very similar to any other distro (which is why benchmarking rarely shows a distro completely crush another, unless there is a regression in one). But fedora (maybe with KDE applied) and Ubuntu are about as different from each other as you can get, and they are both used by more than other distros.
"popularity" can mean different things, but when it comes to raw numbers of users, Fedora and (especially) Ubuntu are more popular than anything else mentioned. Hitting the larges number of possible users makes perfect since. Few people really care if Gentoo gets a 10% advantage over ubuntu on some game or program when the vast majority never touch Gentoo to begin with. But with Fedora added, there will at least be something different to benchmark again, for a more even comparison across Linux overall.
Maybe if you just want raw numbers, Gentoo makes sense. But when looking for user relevance, it isn't a great choice. And if Mir comes along and really puts a huge monkey wrench in the benchmarking department, it would probably be trivial to use a derivative that uses Wayland, like Kubuntu. But it's yet to be seen if that is the case at all.

[chrisb]

would be nice if true, probably bsd model fit more canonical goals than Linux/GPL

Yeah, that is why they release their projects as GPLv3