Why distro choice matters when it comes to game performance

Znurre · 09-22-2010, 07:56 AM

http://jeffhoogland.blogspot.com/201...rformance.html

Some of you might have seen it already.
The results speaks for themselves.

Raideniza · 09-22-2010, 11:52 AM

no.

10char

Angel_de_Combate · 09-22-2010, 11:59 AM

Hmm i'm using Ubuntu 10.04.

Awrath · 09-22-2010, 12:03 PM

Hmm, out of the box, maybe.

It all depends really how you set up your machine, your kernel, your version of mesa, whether you are using a compositing window manager etc... Or at least that is how it is in my experience. I have a low end system as it is, but I am able to get a small performance boost by ensuring I am not using compositing.

But back on topic, I don't really think these tests are accurate enough. Far too many variables involved. Desktop environment (kde/gnome/xfce/lxde/e17 etc..), ubuntu for example uses compiz by default, other DE's have WM that support compisiting which can impact performance, etc...

So out of the box, such differences would be expected, but I think most of us tweak our system so that we can get the best out of it.

Znurre · 09-22-2010, 12:18 PM

No compositing was enabled during this test according to the blog owner.
Also, the NVidia blob is used in all cases, same version, and the NVidia driver does not rely on Mesa.
Maybe the default IO scheduler is playing an important part in these tests though, since Chakra (ArchLinux, more or less) is using noop by default.

NotScias · 09-22-2010, 12:46 PM

What I don't really understand is that Mint which is supposed to be just a sort of Ubuntu + proprietary stuff (flash/codecs/etc...) doesn't suffer from this regression in Ubuntu...

I don't know, I thought that the Linux binaries were the same regardless of the distro. I don't see why there would be performance differences.

Some days ago I switched my desktop from Kubuntu to Archlinux, the FPS in RO are exactly the same.

Znurre · 09-22-2010, 01:01 PM

Quote:

Originally Posted by Shining-Scias

What I don't really understand is that Mint which is supposed to be just a sort of Ubuntu + proprietary stuff (flash/codecs/etc...) doesn't suffer from this regression in Ubuntu...

I don't know, I thought that the Linux binaries were the same regardless of the distro. I don't see why there would be performance differences.

Some days ago I switched my desktop from Kubuntu to Archlinux, the FPS in RO are exactly the same.

In this test he was using the new Mint based on Debian testing, hence why.

Gytha_Ogg · 09-22-2010, 01:08 PM

Make sure you read the legend, though, the graph starts at 2700, not 0.

j4np0l · 09-22-2010, 02:12 PM

I think he should have included Debian, Slackware and Arch in this benchmark and not only distros based on those three.

Also Gentoo maybe, but the performance here depends a lot on the user who performs the installation.

The results are curious, but I would like to know how that difference feels in gaming. Maybe 200 points on that benchmark means 1Fps of difference or maybe 30Fps....I think that kind of information would be nice.

agarn · 09-22-2010, 02:15 PM

This test doesn't list Slackware, therefore it is not a real test

FTA: "As you can see the scores of most of the Linux distros are fairly close (within 3% of the number one), with one exception: Ubuntu.......Ubuntu 10.04 was at the very bottom (over 10% behind Chakra)"

Can a user realistically perceive a 3% difference? I doubt it highly. 10%? possible depending on what is happening at the moment (video lag, etc).

With the wide variety of distros out there I'm surprised there isn't more of a spread between them.

09-22-2010, 07:56 AM	#1
Znurre Marquis Join Date: Jan 2007 Location: RA Posts: 1,897	Why distro choice matters when it comes to game performance http://jeffhoogland.blogspot.com/201...rformance.html Some of you might have seen it already. The results speaks for themselves. __________________ Winning a fight is not what makes you a good player, it's how you do it. http://home.znur.re/screenshot%20201...2011_39_37.jpg

09-22-2010, 12:18 PM	#5
Znurre Marquis Join Date: Jan 2007 Location: RA Posts: 1,897	No compositing was enabled during this test according to the blog owner. Also, the NVidia blob is used in all cases, same version, and the NVidia driver does not rely on Mesa. Maybe the default IO scheduler is playing an important part in these tests though, since Chakra (ArchLinux, more or less) is using noop by default. __________________ Winning a fight is not what makes you a good player, it's how you do it. http://home.znur.re/screenshot%20201...2011_39_37.jpg

09-22-2010, 12:46 PM	#6
NotScias Baron Join Date: Aug 2009 Location: /dev/null Posts: 766	What I don't really understand is that Mint which is supposed to be just a sort of Ubuntu + proprietary stuff (flash/codecs/etc...) doesn't suffer from this regression in Ubuntu... I don't know, I thought that the Linux binaries were the same regardless of the distro. I don't see why there would be performance differences. Some days ago I switched my desktop from Kubuntu to Archlinux, the FPS in RO are exactly the same. __________________ 天 The other realm awaits... Now playing much better games

09-22-2010, 02:12 PM	#9
j4np0l Master Join Date: Dec 2009 Location: Ranting in Montsognir's tabern Posts: 437	I think he should have included Debian, Slackware and Arch in this benchmark and not only distros based on those three. Also Gentoo maybe, but the performance here depends a lot on the user who performs the installation. The results are curious, but I would like to know how that difference feels in gaming. Maybe 200 points on that benchmark means 1Fps of difference or maybe 30Fps....I think that kind of information would be nice. __________________ Janpol - Enano Barbaro - La Ultima Alianza Sos Alsirio, esta es tu casa: http://www.alsius.com.ar/AlsiusForum/ Alsius: always outnumbered, but never unmanned

09-22-2010, 11:52 AM	#2
Raideniza Apprentice Join Date: Dec 2009 Location: Алсиус Posts: 85	no. 10char

09-22-2010, 11:59 AM	#3
Angel_de_Combate Banned Join Date: Jan 2008 Posts: 440	Hmm i'm using Ubuntu 10.04.

09-22-2010, 12:03 PM	#4
Awrath Master Join Date: May 2010 Location: England Posts: 455	Hmm, out of the box, maybe. It all depends really how you set up your machine, your kernel, your version of mesa, whether you are using a compositing window manager etc... Or at least that is how it is in my experience. I have a low end system as it is, but I am able to get a small performance boost by ensuring I am not using compositing. But back on topic, I don't really think these tests are accurate enough. Far too many variables involved. Desktop environment (kde/gnome/xfce/lxde/e17 etc..), ubuntu for example uses compiz by default, other DE's have WM that support compisiting which can impact performance, etc... So out of the box, such differences would be expected, but I think most of us tweak our system so that we can get the best out of it.

09-22-2010, 01:08 PM	#8
Gytha_Ogg Master Join Date: Jan 2009 Location: Valiantly battling at... oops, at Agg save :/ Posts: 253	Make sure you read the legend, though, the graph starts at 2700, not 0.

09-22-2010, 02:15 PM	#10
agarn Pledge Join Date: Apr 2007 Posts: 21	This test doesn't list Slackware, therefore it is not a real test FTA: "As you can see the scores of most of the Linux distros are fairly close (within 3% of the number one), with one exception: Ubuntu.......Ubuntu 10.04 was at the very bottom (over 10% behind Chakra)" Can a user realistically perceive a 3% difference? I doubt it highly. 10%? possible depending on what is happening at the moment (video lag, etc). With the wide variety of distros out there I'm surprised there isn't more of a spread between them.