The Test Setup

One of the great things about Linux is that there are hundreds of distributions available for us to utilize. We are selecting Arch Linux (64-bit) for a few different reasons. The Arch base install is small and does not come packed with pre-configured running services. This will remove any question about what might be running in the background that affects gaming performance.

Arch also has a bleeding-edge implementation of packages. One thing I find in a lot of comments with Linux performance reviews is the standard question, "Did you try the new package that was just released on this nonstandard repository?" Here we will reduce this problem drastically, allowing us to test the latest and greatest Linux has to offer. For the Windows side of testing, we will be using Windows 7 Ultimate, so we'll compare the most up-to-date Linux build with the latest offering from Microsoft. Here are the details of our test system.

Test System
Component Description
Processor Intel Core i7-920 Overclocked to 3.97GHz
(Quad-core + HTT, 45nm, 8MB L3, 4x512KB L2)
RAM OCZ 3x2GB DDR3-1600 (PC3 12800)
Motherboard ASUS Rampage II Extreme
Hard Drives 2 x 74GB Raptors in RAID 0
Video Card EVGA 280 GTX 1GB
Operating Systems Arch Linux (64-bit)
Windows 7 Ultimate 64-bit
Drivers NVIDIA 191.07 (windows)
NVIDIA 190.42 (Linux)

Below is a shortened list of packages relevant for our test on Arch Linux. We will be running on a standard Gnome desktop without all the graphical bells and whistles (i.e. compiz, etc.)

Arch Linux Packages
Package Version
Gnome 2.28.1
Xorg-Server 1.7.1
NVIDIA 190.42
Kernel 2.6.31.5
Wine 1.1.32
Cedega 7.4

Our game selection will be a mixture of genres and release dates. One of the complexities of benchmarking in Linux is the lack of FRAPS or a FRAPS alternative. We have selected games that have built-in benchmarking abilities or at least the option to display FPS. Originally we were looking at testing very recent game releases in our Linux lab. However, after spending many weeks of unsuccessful attempts to get them to work across all three Wine distributions, we fell back to some older release games. We will provide more information on the newer releases tested at the end of this article.

Game Selection
Title Genre Benchmark Method
Eve Online MMORPG (Space/Sci-Fi) Built-in FPS Display
Team Fortress 2 Older FPS Built-in Timedemo
TrackMania Racing Simulation Built-in Benchmark
Unreal Tournament 3 Somewhat Current FPS Built-in Benchmark (War-Serenity)
3DMark06 Benchmark Standard Settings (1280x1024)

Most of the games include in-game benchmarking. We run each benchmark three times and take the average of the three runs for our final results. Eve Online requires the use of the in-game FPS utility. With Eve we found an empty station and recorded FPS exiting the station (180 Seconds). Again we ran these tests three times and use the average of each. Once the benchmarking was completed, I took the time to get in and play the games in order to ensure functionality and find any defects with the gameplay.

Wine Projects: Which Vintage? Linux Gaming Performance
POST A COMMENT

127 Comments

View All Comments

  • ficarra1002 - Friday, February 11, 2011 - link

    Most people don't use Linux because it's free. They get it because its better. You may think its too much work, but that's you. Reply
  • stoggy - Friday, April 29, 2011 - link

    Yea those linux gui server utils are great they make it really easy. then its worth the trouble.

    why did you even read the article if its not worth the bother?

    why do i use linux for a desktop? Because windows cant do what linux can do, period. Even worse is when Windows makes me do what linux does for me.
    Reply
  • zerobug - Saturday, January 02, 2010 - link

    Very good move from Anand, hiring a Linux expert for a better coverage on what's going on in the OSS arena. I welcome this initiative. The present article was a good choice and I'll be checking often for what's following. Good luck and a happy new year to the team and all their readers. Reply
  • ReviveR - Saturday, January 02, 2010 - link

    Comparison is quite unfair, but Linux doesn't actually have to win Windows to become a gaming platform.

    Just today I connected my N900 to TV and played some Star Control 2. It worked really well as cheap console replacement.

    It probably also has more processing power (CPU & GPU) than Nintendo DS or PSP. You can connect it to PC, use some bluetooth based controller like Wiimote etc. I think that when Maemo starts gaining some speed it could be a very nice Linux based gaming platform also.

    If you think how successful Wii has been and then consider that something like N900 can handle bigger resolutions than Wii, offer all the development advantages of Linux etc... The tech is there, all it would need is users realising the possibilities.
    Reply
  • minime - Saturday, January 02, 2010 - link

    I'd like to see more real-life server test (incl. Linux [based] solutions). Something like "Which software stack/solution is able to serve the most users (while being practical meaning, incl. security)" (use-case: webserver, database, java, etc.).

    Then the other way around, keywords: AMD, Intel, PCI-E SSD vs. SATA SSD, GPGPU, etc.
    Reply
  • ChristopherRice - Saturday, January 02, 2010 - link

    In the future, I'd like to do some server articles for sure. Reply
  • minime - Sunday, January 03, 2010 - link

    Cool, thanks! I do believe this is something a lot of readers would like to see and would differentiate from the ordinary IT-blog/news-sauce Reply
  • jediknight - Saturday, January 02, 2010 - link

    I'd be interested in seeing what performance would be like using a VM. Or Is performance so poor as to not even be worth attempting? Reply
  • FluffyTapeworm - Thursday, December 31, 2009 - link

    It would be interesting to see how well running windows + games under vmplayer, virtualbox, etc stacks up. I assume it's likely to be less efficient, but it might still be useful for many older games. Reply
  • vol7ron - Thursday, December 31, 2009 - link

    This article didn't really introduce anything new. All it says is that some people have found a way to manipulate Linux using Wine/etc to play some games.

    That sounds no better than someone issuing a keygen to use an application. Of course, the legalities are better, but you get the point: to only be able to play SOME games at a considerable performance impact does not make this option enticing if you already have Windows.

    All it says is: if you're a Linux user and you want nice gameplay, we recommend you dual boot, as always, because the games you probably really want to play, won't be worth the effort.
    Reply

Log in

Don't have an account? Sign up now