Results 1 to 6 of 6

Thread: SuperTuxKart profile

  1. #1
    Join Date
    Feb 2008
    Location
    Linuxland
    Posts
    5,182

    Default SuperTuxKart profile

    The description is wrong, STK has not used PLIB/SDL in several versions.

    Proposed change:
    Code:
    -This game uses PLIB, SDL, OpenAL, and other technologies.
    +This game uses Irrlicht, OpenAL, and other technologies.
    Also, the profile mode is not a repeatable benchmark, but I bet you knew that already. It is a fresh simulation with random karts and random AI actions.

  2. #2

    Default

    Quote Originally Posted by curaga View Post
    The description is wrong, STK has not used PLIB/SDL in several versions.
    Updated, thanks. http://openbenchmarking.org/test/pts/supertuxkart

    Quote Originally Posted by curaga View Post
    Also, the profile mode is not a repeatable benchmark, but I bet you knew that already. It is a fresh simulation with random karts and random AI actions.
    This test isn't used in many Phoronix articles since it's not too demanding on modern hardware. However, when I have run it I have found the end result to be rather repeatable on modern hardware.

  3. #3
    Join Date
    Feb 2008
    Location
    Linuxland
    Posts
    5,182

    Default

    You're compromising on your own quality by using it next to reproducible benches. The FPS may be close enough (due to cpu bottlenecks), but it's quite obvious to the viewer a completely different race is done each time.

  4. #4
    Join Date
    Feb 2014
    Posts
    15

    Default

    Quote Originally Posted by curaga View Post
    You're compromising on your own quality by using it next to reproducible benches. The FPS may be close enough (due to cpu bottlenecks), but it's quite obvious to the viewer a completely different race is done each time.
    It doesn't hurt to have it in the suite. Not every benchmark is for FPS vs FPS comparison. Sometimes I use these tests just to make sure everything is working as expected & there isn't some form of major regression which is easiest detected in the automation of the suite.

  5. #5
    Join Date
    Feb 2008
    Location
    Linuxland
    Posts
    5,182

    Default

    It does hurt, because it's the only non-reproducible bench in PTS that I know of. Having it there lowers the standards of Phoronix.

    You cannot even regression-test with it, since the AI may decide to get stuck in a corner, leaving you with a wildly different number and you none the wiser.

  6. #6
    Join Date
    Feb 2014
    Posts
    15

    Default

    well, if you go from 400fps to 2fps, you certainly can regression test with it. If you don't like the test don't run it. Just because it doesn't fit your requirements doesn't mean its not useful to others.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •