Benchmarks Of The Gentoo-based Sabayon

[vince]

- init & configuration are grossly over engineered. arch does them much better.

Yeah, and arch doesn't have OpenRC and thus inherently sucks.

[kraftman]

Well, there's something of a sort right... here.

Naturally, I'd like to see Gentoo trash Ubuntu some more, but I'll have to settle for this.

This benchmark is meaningless, because there are different kernel versions used. It doesn't proof Gentoo is faster. It will be better to see generic Gentoo installation vs optimized one.

@Bnolsen

- init & configuration are grossly over engineered. arch does them much better.

This cannot be said about Arch network configuration. It's sometimes a real mess IMHO.

[n0nsense]

This benchmark is meaningless, because there are different kernel versions used. It doesn't proof Gentoo is faster. It will be better to see generic Gentoo installation vs optimized one.
.

There is no such thing as "generic" Gentoo.
It is always "your Gentoo"
Most close to "Generic Gentoo" is Live CD/DVD.
There is also Sabayon But in this case, you have to

• Install Sabayon
• Edit make.conf
• Rebuild the system

And yet, it isn't generic Gentoo.

[clavko]

This benchmark is meaningless, because there are different kernel versions used. It doesn't proof Gentoo is faster. It will be better to see generic Gentoo installation vs optimized one.

I agree. That's why I'm trying to say. For benchmarking to have some sense
one should have as much parameters as it can under control. Gentoo has this
kind of advantage to be a perfect customizable benchmark environment.
Now, I know it's time consuming, that's why I don't expect Phoronix to
jump at it. However... mmm... it sure as hell would be nice

[Apopas]

Compared to having to read extensive documentation to setup a Gentoo system most definitely.

Gentoo needs no setup for this. You just write in /etc/make.conf
+qt -gtk -oss -semantic-desktop etc
The absolute optimization in the simplest way.
It's impossible to do this in OpenSUSE without get mad? Ok in every distro you can do whatever you want, you can even get Ubuntu and transform it to OpenSUSE. Linux is Linux, but the basic matter is to be practical and this is not.

In place upgrades are nothing new in many distros. In openSUSE's case building against factory will suffice. Up until a few months ago I had a system that was upgraded with each release since SuSE 9.1 (which is about 5 years ago)

Oh you mean upgrade from 9.1 to 9.2 etc... I ment to keep 9.1 modern for 5 years, but nevertheless even like that I've yet to see a distro but Gentoo which is reliable about version upgrades. Sometimes it works but usually you have so many problems that you find yourself wondering why you upgraded instead of making a clear installation which would be faster and less problematic. I know this from experience, SuSE was my favourite distro till 2004, before I switched to Gentoo and even now many times I suffer when I decide to upgrade instead of install an Ubuntu system I've installed to a friend.

Not sure about ubuntu but with openSUSE it is very possible which is why it is pleasure to use for JEOS systems.

The matter is what was the reason for that slowness and how could we be sure that only Ubuntu suffers from this? Besides, as far as we've seen in the comparison in this specific site, Ubuntu is from the fastest conventional distros.

[Apopas]

I benchmarked a number of games and then I compiled it again using tailored flags (I believe they were -O3 -march=k8-sse3 -fomit-frame-pointer -mfpmath=sse -ffast-math)

I think -fomit-frame-pointer is enabled by default in X86_64, so is not needed to ad it to gcc flags.

This benchmark is meaningless, because there are different kernel versions used. It doesn't proof Gentoo is faster. It will be better to see generic Gentoo installation vs optimized one.

But one of Gentoo's advantages is to always be modern. If we test same kernel versions etc between Gentoo and Ubuntu or whatever, it means that we kept our Gentoo old, which wouldn't be the case in reality

[XorEaxEax]

I think -fomit-frame-pointer is enabled by default in X86_64, so is not needed to ad it to gcc flags.

Yes you're right, I just add it out of old habit

[ChrisIrwin]

I would have liked to have seen 64-bit (K)Ubuntu thrown in for comparison. I'm assuming the example Ubuntu was the regular i386.

[inode64.com]

I wrote update ebuild of phoronix test suite for Gentoo

http://bugs.gentoo.org/show_bug.cgi?id=299121

[Apopas]

Michael forgot to mention what version of Kubuntu used, but I assume he used the 64 bit one.