Own an ATI X1, HD or NV 7/8 series GPU and have 5-10 minutes? Please run this.

Discussion in 'GPGPU Technology & Programming' started by PeterT, May 29, 2007.

  1. PeterT

    Regular

    Joined:
    May 14, 2002
    Messages:
    702
    Likes Received:
    14
    Location:
    Austria
    Hi enthusiast community, I once again turn to you for your help. If you own one of these cards (I'm especially interested in the new ATI/AMD GPUs of course), use a windows OS, and have a bit of time then please follow the instructions below.

    Please download this: (611 KiB)
    http://peter.metaclassofnil.com/uni/mgbench310507.zip
    It extracts to a folder called "mgbench".
    Then close all GPU-intensive or high-priority CPU apps you may have running and run the .exe contained in the folder. Now, this can take 3 to ? (probably around 10) minutes. After it has finished, it should have generated a file called "GPUMGbench.txt". Either send me that file by email (peter@metaclassofnil.com), or post its contents here, or send it via forum PM. However, please don't forget to mention your graphics card type in detail, and if you use non-standard clock speeds. Driver version would also be nice but is not strictly required.

    Some important points:
    - If you're worried about this being malware, you can see in my posting history that I've already done something like this 2 times before. I've also been a member of this forum for over 5 years.
    - Better not try to use the values reported by this program to compare GPU performance. It's very specific, somewhat overhead-limited and probably buggy.
    - WARNING: If you're not 100% sure that your GPU is stable, save your data or return it to standard clocks before running this. After getting some reports of crashes I found out that this tool is a very good "GPU stability test" - even configurations that make it through hours of 3Dmark loops may well BSoD running this program. I even had one card crash at stock speeds and only be able to complete the run downclocked.

    Thanks for your help!

    [edit]
    I updated the link above to point to the new version, for more information see my post later in this thread.
     
    #1 PeterT, May 29, 2007
    Last edited by a moderator: May 31, 2007
  2. TheAlSpark

    TheAlSpark Moderator
    Moderator Legend

    Joined:
    Feb 29, 2004
    Messages:
    20,814
    Likes Received:
    5,916
    Location:
    ಠ_ಠ
    Q: What is the program doing?
     
  3. PeterT

    Regular

    Joined:
    May 14, 2002
    Messages:
    702
    Likes Received:
    14
    Location:
    Austria
    A: It benchmarks the speed of various components of a GPGPU multigrid method implementation at a number of different problem sizes. The specific algorithm implemented in this benchmark calculates the solution to a 2D Poisson equation (with Dirichlet boundary conditions) using a multigrid algorithm with a Jacobi smoother. A bit of explanation can be found here:
    http://peter.metaclassofnil.com/mastervortrag.pdf
    The benchmark data in that presentation is extremely old and incorrect though, the current version is much more optimized and somewhat more correct. It is part of my Master's thesis which should be coming to a close soon.

    Maybe I should have mentioned that in the first post :D
     
    TheAlSpark likes this.
  4. GMâ„¢

    Newcomer

    Joined:
    Mar 1, 2007
    Messages:
    181
    Likes Received:
    0
    Location:
    Reading, UK.
  5. trinibwoy

    trinibwoy Meh
    Legend

    Joined:
    Mar 17, 2004
    Messages:
    10,436
    Likes Received:
    443
    Location:
    New York
    ATI X1900XT-512 Stock
    Catalyst 7.4
    AthlonX2 2.5Ghz
    Windows XP SP2

    Results:
    Crashed with "atidvag.dll stuck in infinite loop" error :???:
     
  6. AlexV

    AlexV Heteroscedasticitate
    Moderator Veteran

    Joined:
    Mar 15, 2005
    Messages:
    2,528
    Likes Received:
    107
    HD 2900XT 512MB Stock Clocks
    Catalyst 8.37.4.3
    Pentium D 940 3.2Ghz
    Vista X64
    Code:
    
    ------------------
    -- size: 3
    
    BENCHMARK
    Jacobi iteration: 20.3 micros
    Residual calculation: 18.7 micros
    Restriction: fw: 15.6 micros
    Interpolation + add: 18.8 micros
    VCycle: 359 micros
    
    
    BENCHMARK
    Jacobi iteration: 20.3 micros
    Residual calculation: 18.8 micros
    Restriction: fw: 14 micros
    Interpolation + add: 18.8 micros
    VCycle: 390 micros
    
    
    BENCHMARK
    Jacobi iteration: 18.7 micros
    Residual calculation: 20.3 micros
    Restriction: fw: 14.1 micros
    Interpolation + add: 18.7 micros
    VCycle: 360 micros
    
    
    
    ------------------
    -- size: 7
    
    BENCHMARK
    Jacobi iteration: 19.1111 micros
    Residual calculation: 19 micros
    Restriction: fw: 15.6667 micros
    Interpolation + add: 19.1111 micros
    VCycle: 607.778 micros
    
    
    BENCHMARK
    Jacobi iteration: 19.1111 micros
    Residual calculation: 19.1111 micros
    Restriction: fw: 13.8889 micros
    Interpolation + add: 19.1111 micros
    VCycle: 607.778 micros
    
    
    BENCHMARK
    Jacobi iteration: 19.1111 micros
    Residual calculation: 19.1111 micros
    Restriction: fw: 15.6667 micros
    Interpolation + add: 19 micros
    VCycle: 607.778 micros
    
    
    
    ------------------
    -- size: 15
    
    BENCHMARK
    Jacobi iteration: 19.625 micros
    Residual calculation: 19.5 micros
    Restriction: fw: 13.625 micros
    Interpolation + add: 19.625 micros
    VCycle: 820 micros
    
    
    BENCHMARK
    Jacobi iteration: 19.5 micros
    Residual calculation: 19.625 micros
    Restriction: fw: 13.625 micros
    Interpolation + add: 19.5 micros
    VCycle: 840 micros
    
    
    BENCHMARK
    Jacobi iteration: 19.625 micros
    Residual calculation: 17.5 micros
    Restriction: fw: 15.625 micros
    Interpolation + add: 17.625 micros
    VCycle: 840 micros
    
    
    
    ------------------
    -- size: 31
    
    BENCHMARK
    Jacobi iteration: 20.1429 micros
    Residual calculation: 20 micros
    Restriction: fw: 13.4286 micros
    Interpolation + add: 20.1429 micros
    VCycle: 1048.57 micros
    
    
    BENCHMARK
    Jacobi iteration: 20.1429 micros
    Residual calculation: 17.8571 micros
    Restriction: fw: 15.5714 micros
    Interpolation + add: 17.8571 micros
    VCycle: 1094.29 micros
    
    
    BENCHMARK
    Jacobi iteration: 20.1429 micros
    Residual calculation: 20 micros
    Restriction: fw: 13.4286 micros
    Interpolation + add: 20.1429 micros
    VCycle: 1048.57 micros
    
    
    
    ------------------
    -- size: 63
    
    BENCHMARK
    Jacobi iteration: 18.3333 micros
    Residual calculation: 18.1667 micros
    Restriction: fw: 15.6667 micros
    Interpolation + add: 18.1667 micros
    VCycle: 1303.33 micros
    
    
    BENCHMARK
    Jacobi iteration: 20.8333 micros
    Residual calculation: 18.1667 micros
    Restriction: fw: 15.6667 micros
    Interpolation + add: 18.1667 micros
    VCycle: 1328.33 micros
    
    
    BENCHMARK
    Jacobi iteration: 20.8333 micros
    Residual calculation: 18.3333 micros
    Restriction: fw: 15.5 micros
    Interpolation + add: 18.3333 micros
    VCycle: 1301.67 micros
    
    
    
    ------------------
    -- size: 127
    
    BENCHMARK
    Jacobi iteration: 15.6 micros
    Residual calculation: 21.8 micros
    Restriction: fw: 12.6 micros
    Interpolation + add: 21.8 micros
    VCycle: 1500 micros
    
    
    BENCHMARK
    Jacobi iteration: 18.8 micros
    Residual calculation: 18.8 micros
    Restriction: fw: 15.6 micros
    Interpolation + add: 18.8 micros
    VCycle: 1530 micros
    
    
    BENCHMARK
    Jacobi iteration: 22 micros
    Residual calculation: 18.6 micros
    Restriction: fw: 12.6 micros
    Interpolation + add: 18.8 micros
    VCycle: 1562 micros
    
    
    
    ------------------
    -- size: 255
    
    BENCHMARK
    Jacobi iteration: 19.5 micros
    Residual calculation: 19.5 micros
    Restriction: fw: 15.5 micros
    Interpolation + add: 19.5 micros
    VCycle: 1760 micros
    
    
    BENCHMARK
    Jacobi iteration: 19.5 micros
    Residual calculation: 19.5 micros
    Restriction: fw: 15.75 micros
    Interpolation + add: 19.5 micros
    VCycle: 1795 micros
    
    
    BENCHMARK
    Jacobi iteration: 15.5 micros
    Residual calculation: 19.5 micros
    Restriction: fw: 15.75 micros
    Interpolation + add: 19.5 micros
    VCycle: 1797.5 micros
    
    
    
    ------------------
    -- size: 511
    
    BENCHMARK
    Jacobi iteration: 20.6667 micros
    Residual calculation: 21 micros
    Restriction: fw: 10.3333 micros
    Interpolation + add: 20.6667 micros
    VCycle: 2033.33 micros
    
    
    BENCHMARK
    Jacobi iteration: 20.6667 micros
    Residual calculation: 15.6667 micros
    Restriction: fw: 15.6667 micros
    Interpolation + add: 20.6667 micros
    VCycle: 2033.33 micros
    
    
    BENCHMARK
    Jacobi iteration: 20.6667 micros
    Residual calculation: 21 micros
    Restriction: fw: 10.3333 micros
    Interpolation + add: 21 micros
    VCycle: 2030 micros
    
    
    
    ------------------
    -- size: 1023
    
    BENCHMARK
    Jacobi iteration: 15.5 micros
    Residual calculation: 23.5 micros
    Restriction: fw: 15.5 micros
    Interpolation + add: 16 micros
    VCycle: 2340 micros
    
    
    BENCHMARK
    Jacobi iteration: 15.5 micros
    Residual calculation: 23.5 micros
    Restriction: fw: 16 micros
    Interpolation + add: 15.5 micros
    VCycle: 2345 micros
    
    
    BENCHMARK
    Jacobi iteration: 16 micros
    Residual calculation: 23 micros
    Restriction: fw: 8 micros
    Interpolation + add: 23.5 micros
    VCycle: 2265 micros
    
    
    
    ------------------
    -- size: 2047
    
    BENCHMARK
    Jacobi iteration: 32 micros
    Residual calculation: 15 micros
    Restriction: fw: 16 micros
    Interpolation + add: 15 micros
    VCycle: 2660 micros
    
    
    BENCHMARK
    Jacobi iteration: 15 micros
    Residual calculation: 16 micros
    Restriction: fw: 15 micros
    Interpolation + add: 32 micros
    VCycle: 2500 micros
    
    
    BENCHMARK
    Jacobi iteration: 16 micros
    Residual calculation: 15 micros
    Restriction: fw: 16 micros
    Interpolation + add: 16 micros
    VCycle: 2500 micros
    Sorry for posting it like that, but I cannot post attachements.
     
  7. Aerows

    Regular Newcomer

    Joined:
    Nov 19, 2002
    Messages:
    317
    Likes Received:
    6
    Vista 32-bit/Geforce 8600 GTS:

    GL multigrid benchmark 0.5 by Peter Thoman 2007

    DevIL initialized successfully.
    OpenGL initialized successfully.
    ILUT OpenGL mode set successfully.
    GLEW version 1.3.5 initialized successfully.
    Loaded required OpenGL extensions for GLPixelShader.
    Loaded required OpenGL extensions for GLRenderTexture.
    Loaded required OpenGL extensions for GLFilterStep.
    Initialization complete.

    GLImageTexture: DevIL: could not load image.
     
  8. PeterT

    Regular

    Joined:
    May 14, 2002
    Messages:
    702
    Likes Received:
    14
    Location:
    Austria
    Wow. Those scores are utterly amazing if true. I really can't believe them for the 2047 test at least - regardless of how powerful the card is, it can't be that fast... (I'm not accusing you, my tests must be buggy, perhaps it's vista related)

    It would be helpful to get some more HD 2900 results, perhaps running in XP.

    Seems your card is not stable at stock speeds :/ - I really should market this as a stability test.

    That's very strange. Did you extract the zip archive with folders intact?
     
  9. AlexV

    AlexV Heteroscedasticitate
    Moderator Veteran

    Joined:
    Mar 15, 2005
    Messages:
    2,528
    Likes Received:
    107
    I`ve barely looked at the way you`ve set-up your code-if it`s very friendly towards the R600s architecture, it could have some serious oomph...by skimming one of the shaders i`ve noticed quite a number of vec4s being done(skimming as really skimming, so I may not have been paying attention). I simply ran the exe, which means it ran in windowed mode...any command line arguments I should use or anything?
     
  10. Skrying

    Skrying S K R Y I N G
    Veteran

    Joined:
    Jul 8, 2005
    Messages:
    4,815
    Likes Received:
    61
    X1950 Pro
    Core 2 Duo E6300 @ 2.8Ghz
    8.36/Catalyst 7.4
    Windows Vista (32-bit)

    Code:
    GL multigrid benchmark 0.5 by Peter Thoman 2007
    
    DevIL initialized successfully.
    OpenGL initialized successfully.
    ILUT OpenGL mode set successfully.
    GLEW version 1.3.5 initialized successfully.
    Loaded required OpenGL extensions for GLPixelShader.
    Loaded required OpenGL extensions for GLRenderTexture.
    Loaded required OpenGL extensions for GLFilterStep.
    Initialization complete.
    
    
    
    ------------------
    -- size: 3
    
    BENCHMARK
    Jacobi iteration: 11 micros
    Residual calculation: 11.2 micros
    Restriction: fw: 8.2 micros
    Interpolation + add: 11 micros
    VCycle: 226 micros
    
    
    BENCHMARK
    Jacobi iteration: 11 micros
    Residual calculation: 11.1 micros
    Restriction: fw: 8.2 micros
    Interpolation + add: 11.1 micros
    VCycle: 223 micros
    
    
    BENCHMARK
    Jacobi iteration: 11.1 micros
    Residual calculation: 11.2 micros
    Restriction: fw: 8.2 micros
    Interpolation + add: 10.9 micros
    VCycle: 223 micros
    
    
    
    ------------------
    -- size: 7
    
    BENCHMARK
    Jacobi iteration: 11.3333 micros
    Residual calculation: 11.2222 micros
    Restriction: fw: 8.11111 micros
    Interpolation + add: 11.2222 micros
    VCycle: 364.444 micros
    
    
    BENCHMARK
    Jacobi iteration: 11.2222 micros
    Residual calculation: 11.2222 micros
    Restriction: fw: 8.22222 micros
    Interpolation + add: 11 micros
    VCycle: 366.667 micros
    
    
    BENCHMARK
    Jacobi iteration: 11.2222 micros
    Residual calculation: 11 micros
    Restriction: fw: 8.22222 micros
    Interpolation + add: 11.2222 micros
    VCycle: 364.444 micros
    
    
    
    ------------------
    -- size: 15
    
    BENCHMARK
    Jacobi iteration: 11.125 micros
    Residual calculation: 11.125 micros
    Restriction: fw: 8.125 micros
    Interpolation + add: 11.25 micros
    VCycle: 503.75 micros
    
    
    BENCHMARK
    Jacobi iteration: 10.875 micros
    Residual calculation: 11.125 micros
    Restriction: fw: 8.125 micros
    Interpolation + add: 11.125 micros
    VCycle: 500 micros
    
    
    BENCHMARK
    Jacobi iteration: 11 micros
    Residual calculation: 11 micros
    Restriction: fw: 8.25 micros
    Interpolation + add: 11.125 micros
    VCycle: 503.75 micros
    
    
    
    ------------------
    -- size: 31
    
    BENCHMARK
    Jacobi iteration: 11.4286 micros
    Residual calculation: 11.2857 micros
    Restriction: fw: 8.14286 micros
    Interpolation + add: 11.2857 micros
    VCycle: 644.286 micros
    
    
    BENCHMARK
    Jacobi iteration: 11.2857 micros
    Residual calculation: 11.1429 micros
    Restriction: fw: 8.42857 micros
    Interpolation + add: 11.1429 micros
    VCycle: 645.714 micros
    
    
    BENCHMARK
    Jacobi iteration: 11.1429 micros
    Residual calculation: 11.1429 micros
    Restriction: fw: 8.42857 micros
    Interpolation + add: 11.1429 micros
    VCycle: 644.286 micros
    
    
    
    ------------------
    -- size: 63
    
    BENCHMARK
    Jacobi iteration: 11.5 micros
    Residual calculation: 11.5 micros
    Restriction: fw: 8.5 micros
    Interpolation + add: 11.5 micros
    VCycle: 806.667 micros
    
    
    BENCHMARK
    Jacobi iteration: 11.5 micros
    Residual calculation: 11.5 micros
    Restriction: fw: 8.5 micros
    Interpolation + add: 11.6667 micros
    VCycle: 800 micros
    
    
    BENCHMARK
    Jacobi iteration: 11.8333 micros
    Residual calculation: 11.6667 micros
    Restriction: fw: 8.5 micros
    Interpolation + add: 11.8333 micros
    VCycle: 798.333 micros
    
    
    
    ------------------
    -- size: 127
    
    BENCHMARK
    Jacobi iteration: 11 micros
    Residual calculation: 11.4 micros
    Restriction: fw: 8 micros
    Interpolation + add: 11 micros
    VCycle: 928 micros
    
    
    BENCHMARK
    Jacobi iteration: 10.8 micros
    Residual calculation: 11.4 micros
    Restriction: fw: 8 micros
    Interpolation + add: 11.2 micros
    VCycle: 926 micros
    
    
    BENCHMARK
    Jacobi iteration: 11.2 micros
    Residual calculation: 11 micros
    Restriction: fw: 8 micros
    Interpolation + add: 11.2 micros
    VCycle: 922 micros
    
    
    
    ------------------
    -- size: 255
    
    BENCHMARK
    Jacobi iteration: 12 micros
    Residual calculation: 11.75 micros
    Restriction: fw: 8.5 micros
    Interpolation + add: 11.75 micros
    VCycle: 1082.5 micros
    
    
    BENCHMARK
    Jacobi iteration: 11.25 micros
    Residual calculation: 11.25 micros
    Restriction: fw: 8.25 micros
    Interpolation + add: 11.5 micros
    VCycle: 1077.5 micros
    
    
    BENCHMARK
    Jacobi iteration: 11.5 micros
    Residual calculation: 11.25 micros
    Restriction: fw: 8.25 micros
    Interpolation + add: 11.25 micros
    VCycle: 1077.5 micros
    
    
    
    ------------------
    -- size: 511
    
    BENCHMARK
    Jacobi iteration: 13 micros
    Residual calculation: 13 micros
    Restriction: fw: 9.33333 micros
    Interpolation + add: 11.3333 micros
    VCycle: 1216.67 micros
    
    
    BENCHMARK
    Jacobi iteration: 11 micros
    Residual calculation: 11.3333 micros
    Restriction: fw: 8 micros
    Interpolation + add: 11.6667 micros
    VCycle: 1220 micros
    
    
    BENCHMARK
    Jacobi iteration: 11.3333 micros
    Residual calculation: 11.3333 micros
    Restriction: fw: 8 micros
    Interpolation + add: 11.3333 micros
    VCycle: 1210 micros
    
    
    
    ------------------
    -- size: 1023
    
    BENCHMARK
    Jacobi iteration: 11 micros
    Residual calculation: 11.5 micros
    Restriction: fw: 7.5 micros
    Interpolation + add: 11.5 micros
    VCycle: 1350 micros
    
    
    BENCHMARK
    Jacobi iteration: 11 micros
    Residual calculation: 11.5 micros
    Restriction: fw: 8 micros
    Interpolation + add: 11.5 micros
    VCycle: 1355 micros
    
    
    BENCHMARK
    Jacobi iteration: 11.5 micros
    Residual calculation: 12 micros
    Restriction: fw: 8.5 micros
    Interpolation + add: 11.5 micros
    VCycle: 1380 micros
    
    
    
    ------------------
    -- size: 2047
    
    BENCHMARK
    Jacobi iteration: 13 micros
    Residual calculation: 16 micros
    Restriction: fw: 11 micros
    Interpolation + add: 12 micros
    VCycle: 1580 micros
    
    
    BENCHMARK
    Jacobi iteration: 11 micros
    Residual calculation: 11 micros
    Restriction: fw: 8 micros
    Interpolation + add: 12 micros
    VCycle: 1490 micros
    
    
    BENCHMARK
    Jacobi iteration: 12 micros
    Residual calculation: 11 micros
    Restriction: fw: 9 micros
    Interpolation + add: 10 micros
    VCycle: 1550 micros
    
     
  11. AnarchX

    Veteran

    Joined:
    Apr 19, 2007
    Messages:
    1,559
    Likes Received:
    34
    P4 3GHz
    6800GT FW158.27
    Win XP SP2


    Code:
    GL multigrid benchmark 0.5 by Peter Thoman 2007
    
    DevIL initialized successfully.
    OpenGL initialized successfully.
    ILUT OpenGL mode set successfully.
    GLEW version 1.3.5 initialized successfully.
    Loaded required OpenGL extensions for GLPixelShader.
    Loaded required OpenGL extensions for GLRenderTexture.
    Loaded required OpenGL extensions for GLFilterStep.
    Initialization complete.
    
    
    
    ------------------
    -- size: 3
    
    BENCHMARK
    Jacobi iteration: 18.7 micros
    Residual calculation: 18.8 micros
    Restriction: fw: 14.1 micros
    Interpolation + add: 14 micros
    VCycle: 828 micros
    
    
    BENCHMARK
    Jacobi iteration: 12.5 micros
    Residual calculation: 12.5 micros
    Restriction: fw: 9.4 micros
    Interpolation + add: 9.4 micros
    VCycle: 797 micros
    
    
    BENCHMARK
    Jacobi iteration: 10.9 micros
    Residual calculation: 12.5 micros
    Restriction: fw: 9.4 micros
    Interpolation + add: 9.4 micros
    VCycle: 797 micros
    
    
    
    ------------------
    -- size: 7
    
    BENCHMARK
    Jacobi iteration: 12.2222 micros
    Residual calculation: 12.1111 micros
    Restriction: fw: 10.4444 micros
    Interpolation + add: 10.4444 micros
    VCycle: 1318.89 micros
    
    
    BENCHMARK
    Jacobi iteration: 12.1111 micros
    Residual calculation: 13.8889 micros
    Restriction: fw: 8.66667 micros
    Interpolation + add: 10.4444 micros
    VCycle: 1441.11 micros
    
    
    BENCHMARK
    Jacobi iteration: 12.2222 micros
    Residual calculation: 12.1111 micros
    Restriction: fw: 8.66667 micros
    Interpolation + add: 10.4444 micros
    VCycle: 1354.44 micros
    
    
    
    ------------------
    -- size: 15
    
    BENCHMARK
    Jacobi iteration: 13.625 micros
    Residual calculation: 13.625 micros
    Restriction: fw: 9.75 micros
    Interpolation + add: 9.875 micros
    VCycle: 1875 micros
    
    
    BENCHMARK
    Jacobi iteration: 13.625 micros
    Residual calculation: 11.75 micros
    Restriction: fw: 9.75 micros
    Interpolation + add: 11.75 micros
    VCycle: 2011.25 micros
    
    
    BENCHMARK
    Jacobi iteration: 13.75 micros
    Residual calculation: 13.625 micros
    Restriction: fw: 9.75 micros
    Interpolation + add: 9.75 micros
    VCycle: 1875 micros
    
    
    
    ------------------
    -- size: 31
    
    BENCHMARK
    Jacobi iteration: 24.5714 micros
    Residual calculation: 22.2857 micros
    Restriction: fw: 13.4286 micros
    Interpolation + add: 15.5714 micros
    VCycle: 2678.57 micros
    
    
    BENCHMARK
    Jacobi iteration: 24.5714 micros
    Residual calculation: 22.2857 micros
    Restriction: fw: 13.4286 micros
    Interpolation + add: 15.5714 micros
    VCycle: 2522.86 micros
    
    
    BENCHMARK
    Jacobi iteration: 22.2857 micros
    Residual calculation: 22.4286 micros
    Restriction: fw: 13.2857 micros
    Interpolation + add: 15.7143 micros
    VCycle: 2454.29 micros
    
    
    
    ------------------
    -- size: 63
    
    BENCHMARK
    Jacobi iteration: 60 micros
    Residual calculation: 57.1667 micros
    Restriction: fw: 34 micros
    Interpolation + add: 39 micros
    VCycle: 3073.33 micros
    
    
    BENCHMARK
    Jacobi iteration: 60 micros
    Residual calculation: 59.8333 micros
    Restriction: fw: 33.8333 micros
    Interpolation + add: 36.5 micros
    VCycle: 3125 micros
    
    
    BENCHMARK
    Jacobi iteration: 59.8333 micros
    Residual calculation: 59.8333 micros
    Restriction: fw: 34 micros
    Interpolation + add: 36.3333 micros
    VCycle: 3125 micros
    
    
    
    ------------------
    -- size: 127
    
    BENCHMARK
    Jacobi iteration: 253 micros
    Residual calculation: 253.2 micros
    Restriction: fw: 171.8 micros
    Interpolation + add: 128.2 micros
    VCycle: 3938 micros
    
    
    BENCHMARK
    Jacobi iteration: 253.2 micros
    Residual calculation: 253.2 micros
    Restriction: fw: 168.6 micros
    Interpolation + add: 128.2 micros
    VCycle: 3938 micros
    
    
    BENCHMARK
    Jacobi iteration: 253 micros
    Residual calculation: 250 micros
    Restriction: fw: 172 micros
    Interpolation + add: 128 micros
    VCycle: 3908 micros
    
    
    
    ------------------
    -- size: 255
    
    BENCHMARK
    Jacobi iteration: 679.75 micros
    Residual calculation: 675.75 micros
    Restriction: fw: 492.25 micros
    Interpolation + add: 445.25 micros
    VCycle: 8437.5 micros
    
    
    BENCHMARK
    Jacobi iteration: 683.5 micros
    Residual calculation: 675.75 micros
    Restriction: fw: 488.25 micros
    Interpolation + add: 445.25 micros
    VCycle: 8400 micros
    
    
    BENCHMARK
    Jacobi iteration: 683.5 micros
    Residual calculation: 672 micros
    Restriction: fw: 492 micros
    Interpolation + add: 445.5 micros
    VCycle: 8397.5 micros
    
    
    
    ------------------
    -- size: 511
    
    BENCHMARK
    Jacobi iteration: 2859.33 micros
    Residual calculation: 2859.33 micros
    Restriction: fw: 2265.67 micros
    Interpolation + add: 2010.33 micros
    VCycle: 33073.3 micros
    
    
    BENCHMARK
    Jacobi iteration: 2844 micros
    Residual calculation: 2859.33 micros
    Restriction: fw: 2270.67 micros
    Interpolation + add: 1979.33 micros
    VCycle: 35520 micros
    
    
    BENCHMARK
    Jacobi iteration: 3177.33 micros
    Residual calculation: 3375 micros
    Restriction: fw: 2723.67 micros
    Interpolation + add: 2088.67 micros
    VCycle: 36146.7 micros
    
    
    
    ------------------
    -- size: 1023
    
    BENCHMARK
    Jacobi iteration: 14218.5 micros
    Residual calculation: 14515.5 micros
    Restriction: fw: 12484.5 micros
    Interpolation + add: 11429.5 micros
    VCycle: 165470 micros
    
    
    BENCHMARK
    Jacobi iteration: 14101.5 micros
    Residual calculation: 14500 micros
    Restriction: fw: 12539 micros
    Interpolation + add: 11476.5 micros
    VCycle: 168360 micros
    
    
    BENCHMARK
    Jacobi iteration: 14164 micros
    Residual calculation: 14445.5 micros
    Restriction: fw: 12508 micros
    Interpolation + add: 11468.5 micros
    VCycle: 165315 micros
    
    
    
    ------------------
    -- size: 2047
    GLFilterStep -- Error doing makeCurrent on target.
    size 2047 did not run?
     
  12. AlexV

    AlexV Heteroscedasticitate
    Moderator Veteran

    Joined:
    Mar 15, 2005
    Messages:
    2,528
    Likes Received:
    107
    Seeing this, I`d think there`s either a bug in your code or something the Catalysts foobar in order to achieve such performance. It should also be mentioned that, at least for me, the test was quite short, dunno if it reached 3 minutes.
     
  13. Skrying

    Skrying S K R Y I N G
    Veteran

    Joined:
    Jul 8, 2005
    Messages:
    4,815
    Likes Received:
    61
    Oh yes, I should mention it was about 30 seconds for me on the test.
     
  14. PeterT

    Regular

    Joined:
    May 14, 2002
    Messages:
    702
    Likes Received:
    14
    Location:
    Austria
    That's normal, may run out of memory on some cards.

    Looking at Skrying's results, who also runs an ATI card on Vista, it's clear that there's something very wrong going on with that combination. There's simply no way these results can be right. Rule of thumb: if any of "Jacobi iteration", "Residual calculation", "Restriction" or "Interpolation + add" report less than 100 microseconds at size 2047 (or even 1023) the measurements are not being taken correctly.

    Here are my 8800 gts results at those sizes for comparison. That's about the range those numbers "should" fall in.
    Code:
    ------------------
    -- size: 1023
    
    BENCHMARK
    Jacobi iteration: 1320.5 micros
    Residual calculation: 1265.5 micros
    Restriction: fw: 1250 micros
    Interpolation + add: 961 micros
    VCycle: 23750 micros
    
    
    BENCHMARK
    Jacobi iteration: 1344 micros
    Residual calculation: 1234 micros
    Restriction: fw: 1250 micros
    Interpolation + add: 961 micros
    VCycle: 23750 micros
    
    
    BENCHMARK
    Jacobi iteration: 1351.5 micros
    Residual calculation: 1234 micros
    Restriction: fw: 1242.5 micros
    Interpolation + add: 961 micros
    VCycle: 23750 micros
    
    
    
    ------------------
    -- size: 2047
    
    BENCHMARK
    Jacobi iteration: 4456 micros
    Residual calculation: 4734 micros
    Restriction: fw: 5594 micros
    Interpolation + add: 4922 micros
    VCycle: 139690 micros
    
    
    BENCHMARK
    Jacobi iteration: 4509 micros
    Residual calculation: 4703 micros
    Restriction: fw: 5688 micros
    Interpolation + add: 4875 micros
    VCycle: 140000 micros
    
    
    BENCHMARK
    Jacobi iteration: 4625 micros
    Residual calculation: 4750 micros
    Restriction: fw: 5656 micros
    Interpolation + add: 4844 micros
    VCycle: 140160 micros
    I really need to investigate what's going on with ATI in Vista, but that's somewhat challenging if one has neither at hand.

    [edit]
    Indeed. It would be great if someone running ATI on XP could post, to see if it's a general Catalyst issue or a Catalyst/Vista issue.
     
  15. Dave Baumann

    Dave Baumann Gamerscore Wh...
    Moderator Legend

    Joined:
    Jan 29, 2002
    Messages:
    14,079
    Likes Received:
    648
    Location:
    O Canada!
    With regards to clockspeeds, I'm not convinced that this would be operating at "3D speeds" on ATI Radeon boards.
     
  16. PeterT

    Regular

    Joined:
    May 14, 2002
    Messages:
    702
    Likes Received:
    14
    Location:
    Austria
    Well, currently it seems to operate at "well beyond lightspeed" on ATI cards in Vista, so that shouldn't be a problem ;)

    But seriously, I hadn't considered that. What triggers "3D speeds"?
     
  17. Dave Baumann

    Dave Baumann Gamerscore Wh...
    Moderator Legend

    Joined:
    Jan 29, 2002
    Messages:
    14,079
    Likes Received:
    648
    Location:
    O Canada!
    Full screen rendering.
     
  18. Aerows

    Regular Newcomer

    Joined:
    Nov 19, 2002
    Messages:
    317
    Likes Received:
    6
    I would put this in a code box if I knew how...If someone could tell me how to do this I will go back and edit it. The following are my results with my 8800 GTS. I also have results with a Geforce 8600 GTS. I will wait, though, if someone would be so kind as to explain how to fix it into the scrolling box so that it won't be so irritating to read ;).

    AMD 4000+ X2 Brisbane @ 2.83
    G.SKILL 4-4-4-12 @ 704mhz
    Geforce 8800 GTS 640MB @ 525/830
    Forceware 158.22/ Windows XP SP2
    **************************************
    Code:
    GL multigrid benchmark 0.5 by Peter Thoman 2007
    
    DevIL initialized successfully.
    OpenGL initialized successfully.
    ILUT OpenGL mode set successfully.
    GLEW version 1.3.5 initialized successfully.
    Loaded required OpenGL extensions for GLPixelShader.
    Loaded required OpenGL extensions for GLRenderTexture.
    Loaded required OpenGL extensions for GLFilterStep.
    Initialization complete.
    
    
    
    ------------------
    -- size: 3
    
    BENCHMARK
    Jacobi iteration: 39.1 micros
    Residual calculation: 35.9 micros
    Restriction: fw: 39.1 micros
    Interpolation + add: 39.1 micros
    VCycle: 1515 micros
    
    
    BENCHMARK
    Jacobi iteration: 37.5 micros
    Residual calculation: 39.1 micros
    Restriction: fw: 37.5 micros
    Interpolation + add: 39 micros
    VCycle: 1532 micros
    
    
    BENCHMARK
    Jacobi iteration: 37.5 micros
    Residual calculation: 39.1 micros
    Restriction: fw: 37.5 micros
    Interpolation + add: 31.2 micros
    VCycle: 1547 micros
    
    
    
    ------------------
    -- size: 7
    
    BENCHMARK
    Jacobi iteration: 39.8889 micros
    Residual calculation: 29.5556 micros
    Restriction: fw: 38.2222 micros
    Interpolation + add: 38.1111 micros
    VCycle: 2483.33 micros
    
    
    BENCHMARK
    Jacobi iteration: 36.4444 micros
    Residual calculation: 38.1111 micros
    Restriction: fw: 38.2222 micros
    Interpolation + add: 38.2222 micros
    VCycle: 2465.56 micros
    
    
    BENCHMARK
    Jacobi iteration: 38.2222 micros
    Residual calculation: 38.1111 micros
    Restriction: fw: 36.4444 micros
    Interpolation + add: 31.3333 micros
    VCycle: 2500 micros
    
    
    
    ------------------
    -- size: 15
    
    BENCHMARK
    Jacobi iteration: 33.25 micros
    Residual calculation: 39 micros
    Restriction: fw: 37.125 micros
    Interpolation + add: 39.125 micros
    VCycle: 3417.5 micros
    
    
    BENCHMARK
    Jacobi iteration: 37.125 micros
    Residual calculation: 39.125 micros
    Restriction: fw: 39 micros
    Interpolation + add: 37.125 micros
    VCycle: 3437.5 micros
    
    
    BENCHMARK
    Jacobi iteration: 29.375 micros
    Residual calculation: 39 micros
    Restriction: fw: 37.125 micros
    Interpolation + add: 39 micros
    VCycle: 3437.5 micros
    
    
    
    ------------------
    -- size: 31
    
    BENCHMARK
    Jacobi iteration: 38 micros
    Residual calculation: 37.8571 micros
    Restriction: fw: 38 micros
    Interpolation + add: 38 micros
    VCycle: 4374.29 micros
    
    
    BENCHMARK
    Jacobi iteration: 31.2857 micros
    Residual calculation: 40.1429 micros
    Restriction: fw: 38 micros
    Interpolation + add: 37.8571 micros
    VCycle: 4375.71 micros
    
    
    BENCHMARK
    Jacobi iteration: 38 micros
    Residual calculation: 38 micros
    Restriction: fw: 37.8571 micros
    Interpolation + add: 40.2857 micros
    VCycle: 4374.29 micros
    
    
    
    ------------------
    -- size: 63
    
    BENCHMARK
    Jacobi iteration: 39.1667 micros
    Residual calculation: 39 micros
    Restriction: fw: 36.5 micros
    Interpolation + add: 36.5 micros
    VCycle: 5363.33 micros
    
    
    BENCHMARK
    Jacobi iteration: 36.5 micros
    Residual calculation: 39 micros
    Restriction: fw: 39.1667 micros
    Interpolation + add: 36.3333 micros
    VCycle: 5338.33 micros
    
    
    BENCHMARK
    Jacobi iteration: 39.1667 micros
    Residual calculation: 39 micros
    Restriction: fw: 36.5 micros
    Interpolation + add: 39 micros
    VCycle: 5365 micros
    
    
    
    ------------------
    -- size: 127
    
    BENCHMARK
    Jacobi iteration: 37.4 micros
    Residual calculation: 37.6 micros
    Restriction: fw: 40.6 micros
    Interpolation + add: 37.4 micros
    VCycle: 6312 micros
    
    
    BENCHMARK
    Jacobi iteration: 12.6 micros
    Residual calculation: 31.2 micros
    Restriction: fw: 37.6 micros
    Interpolation + add: 37.4 micros
    VCycle: 6312 micros
    
    
    BENCHMARK
    Jacobi iteration: 37.4 micros
    Residual calculation: 37.6 micros
    Restriction: fw: 37.4 micros
    Interpolation + add: 40.6 micros
    VCycle: 6282 micros
    
    
    
    ------------------
    -- size: 255
    
    BENCHMARK
    Jacobi iteration: 89.75 micros
    Residual calculation: 78 micros
    Restriction: fw: 74.25 micros
    Interpolation + add: 47 micros
    VCycle: 7617.5 micros
    
    
    BENCHMARK
    Jacobi iteration: 85.75 micros
    Residual calculation: 78.25 micros
    Restriction: fw: 70.25 micros
    Interpolation + add: 47 micros
    VCycle: 7617.5 micros
    
    
    BENCHMARK
    Jacobi iteration: 86 micros
    Residual calculation: 74.25 micros
    Restriction: fw: 70.25 micros
    Interpolation + add: 54.75 micros
    VCycle: 7617.5 micros
    
    
    
    ------------------
    -- size: 511
    
    BENCHMARK
    Jacobi iteration: 323 micros
    Residual calculation: 307.333 micros
    Restriction: fw: 312.333 micros
    Interpolation + add: 192.667 micros
    VCycle: 10106.7 micros
    
    
    BENCHMARK
    Jacobi iteration: 323 micros
    Residual calculation: 302 micros
    Restriction: fw: 317.667 micros
    Interpolation + add: 187.667 micros
    VCycle: 10156.7 micros
    
    
    BENCHMARK
    Jacobi iteration: 317.667 micros
    Residual calculation: 317.667 micros
    Restriction: fw: 312.667 micros
    Interpolation + add: 187.333 micros
    VCycle: 10416.7 micros
    
    
    
    ------------------
    -- size: 1023
    
    BENCHMARK
    Jacobi iteration: 1211 micros
    Residual calculation: 1187.5 micros
    Restriction: fw: 1328.5 micros
    Interpolation + add: 843.5 micros
    VCycle: 24765 micros
    
    
    BENCHMARK
    Jacobi iteration: 1273.5 micros
    Residual calculation: 1187.5 micros
    Restriction: fw: 1320.5 micros
    Interpolation + add: 836 micros
    VCycle: 24840 micros
    
    
    BENCHMARK
    Jacobi iteration: 1242.5 micros
    Residual calculation: 1210.5 micros
    Restriction: fw: 1312.5 micros
    Interpolation + add: 852 micros
    VCycle: 24765 micros
    
    
    
    ------------------
    -- size: 2047
    
    BENCHMARK
    Jacobi iteration: 547 micros
    Residual calculation: 3844 micros
    Restriction: fw: 5687 micros
    Interpolation + add: 4453 micros
    VCycle: 154220 micros
    
    
    BENCHMARK
    Jacobi iteration: 594 micros
    Residual calculation: 3844 micros
    Restriction: fw: 5734 micros
    Interpolation + add: 4359 micros
    VCycle: 155160 micros
    
    
    BENCHMARK
    Jacobi iteration: 578 micros
    Residual calculation: 3828 micros
    Restriction: fw: 5781 micros
    Interpolation + add: 4344 micros
    VCycle: 154690 micros
    
    
    ******************************

    I also have a result for Vista using an 8600 GTS 720/2100 on an X2 @ ~2.4, but will wait until someone shows me how to stick it in a code box ;).
     
    #18 Aerows, May 29, 2007
    Last edited by a moderator: May 29, 2007
  19. PeterT

    Regular

    Joined:
    May 14, 2002
    Messages:
    702
    Likes Received:
    14
    Location:
    Austria
    Put (code) and (/code) around it, but with [ and ] instead of ( and ). Thanks for the results BTW. All NV results so far fall within expectations. But that's not very surprising given that it was developed mostly on a 8800.

    I'd still like some ATI on XP results, to see if the same issue exists there.
     
  20. neliz

    neliz GIGABYTE Man
    Veteran

    Joined:
    Mar 30, 2005
    Messages:
    4,904
    Likes Received:
    23
    Location:
    In the know
    type [ code ] and [ / code ] around it ;)
     
Loading...

Share This Page

  • About Us

    Beyond3D has been around for over a decade and prides itself on being the best place on the web for in-depth, technically-driven discussion and analysis of 3D graphics hardware. If you love pixels and transistors, you've come to the right place!

    Beyond3D is proudly published by GPU Tools Ltd.
Loading...