9-Way April 2014 Linux Distribution Benchmarks

[phoronix]

Phoronix: 9-Way April 2014 Linux Distribution Benchmarks

Building upon earlier Ubuntu 14.04 LTS benchmarks and tests of Red Hat Enterprise Linux 7, here's a nine-way Linux distribution comparison done from the same hardware.

http://www.phoronix.com/vr.php?view=20286

[christian.a]

I ask myself, if it is really representable to compare distributions which has been designed for diefferent workloads.

Debian will be a server Distro, i guess. Ubuntu instead is optimized for desktop usage. These distros
alone will/must perform differently on database and compiling workloads.

so, how can i read the test results if i want to know if ubuntu or debian is better for desktop usage?

[ArthurBorsboom]

I ask myself, if it is really representable to compare distributions which has been designed for diefferent workloads.

Debian will be a server Distro, i guess. Ubuntu instead is optimized for desktop usage. These distros
alone will/must perform differently on database and compiling workloads.

so, how can i read the test results if i want to know if ubuntu or debian is better for desktop usage?

You are (kind of) writing, what I am missing in this article: a conclusion or summary.

[frign]

Gentoo FTW

What about GENTOO? It'd pwn them all!

[chrisb]

I ask myself, if it is really representable to compare distributions which has been designed for diefferent workloads.

Debian will be a server Distro, i guess. Ubuntu instead is optimized for desktop usage. These distros
alone will/must perform differently on database and compiling workloads.

so, how can i read the test results if i want to know if ubuntu or debian is better for desktop usage?

They are comparable, the software is basically the same. A lot of the performance benchmarks depend mostly on the kernel. The 3D benchmarks depend on the kernel, Xorg and drivers. Many distributions just compile the upstream source with no, or very few, performance impacting modifications.

There are plenty of people running Debian desktops, and there are plenty of people running Ubuntu on servers. Same with all of the distributions. Steam is probably the only real hyper-specialised distribution.

[christian.a]

They are comparable, the software is basically the same. A lot of the performance benchmarks depend mostly on the kernel. The 3D benchmarks depend on the kernel, Xorg and drivers. Many distributions just compile the upstream source with no, or very few, performance impacting modifications.

There are plenty of people running Debian desktops, and there are plenty of people running Ubuntu on servers. Same with all of the distributions. Steam is probably the only real hyper-specialised distribution.

They use the same versions of the kernel, maybe. But, the system configuration they apply is different. For example:
Debain uses CFQ as IO Scheduler, Ubuntu uses deadline. This fact does a huge difference, i think. When i try to use
debian as a Desktop OS it will be less responsive during work than Ubuntu will. Vice versa is Ubuntu not so good as server
OS. There is a ubuntu server which could be benchmarked against debian. Even the running deamons could make a difference
i could not ignore. Those desktiop environments will eat ram, on ubuntu server there is no desktop environment running and so
not need to start bluetooth or Networkmanager. Steam os i tend to be very user input agnostic, debian will not.

The point is, i could not extract the essential information this comparison will/should give to me or anyone.

[clavko]

What about GENTOO? It'd pwn them all!

Sssshhh! Gentoo is a forbidden distro on Phoronix. It actually doesn't exist. There are
other distros that are 'as good as Gentoo' - namely Sabayon and Calculate, and these
were benchmarked on Phoronix, showing that Gentoo is actually not faster.

[/sarcasm]

[frign]

You're wrong

Sssshhh! Gentoo is a forbidden distro on Phoronix. It actually doesn't exist. There are
other distros that are 'as good as Gentoo' - namely Sabayon and Calculate, and these
were benchmarked on Phoronix, showing that Gentoo is actually not faster.

[/sarcasm]

To be fair, given Gentoo offers a wide range of configurations, it's hard to benchmark it.
Now, concerning speed, you can't compare Gentoo with Sabayon and Calculate, given both are based on precompiled packages.

Now, talking about compiling software yourself, I'm not talking about sick CCFLAGS, but the possibility to not compile stuff in you don't want.

That's the reason why Gentoo is in fact faster. It's a mathematical certainty!

[clavko]

To be fair, given Gentoo offers a wide range of configurations, it's hard to benchmark it.
Now, concerning speed, you can't compare Gentoo with Sabayon and Calculate, given both are based on precompiled packages.
Now, talking about compiling software yourself, I'm not talking about sick CCFLAGS, but the possibility to not compile stuff in you don't want.
That's the reason why Gentoo is in fact faster. It's a mathematical certainty!

Yeah, there is a range of configurations, but I think most Gentoo-ists would be
happy with a machine tailored kernel and a (-O2 -march=native) system/world
recompiled setup. I'd say that's about minimum a typical Gentoo install is.
And I'm actually not so sure about the speedup. I'd say 10% max.

[frign]

Yeah, there is a range of configurations, but I think most Gentoo-ists would be
happy with a machine tailored kernel and a (-O2 -march=native) system/world
recompiled setup. I'd say that's about minimum a typical Gentoo install is.
And I'm actually not so sure about the speedup. I'd say 10% max.

CFLAGS don't speed up that much! It's all about the USE-flags instead, which can bring around 5-10% improvement.
Of course, if you benchmark Gentoo against Ubuntu and include the same features (read: all available) into a given program to make it an equal run, there won't be much of a difference.