Announcement

Collapse
No announcement yet.

RFO 3DMax Rendering Benchmark

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    RFO 3DMax Rendering Benchmark

    Hello everyone Thanks Munkholm for giving me the opportunity to start this thread

    Klaus had an wonderful idea to have a RFO 3dmax scene(2012 version), that can be used as a rendering benchmark scene, with some of the most interesting 3dmax rendering engines from the market. So he asked me to help him with this nice idea.

    Here is a short info about what we did here:

    To keep it short: i`ve setup a 3dmax scene that has the simple OOTB Revit sample scene imported as a fbx. Some of the materials have been changed (in terms of aspect), some vegetation has been added, also a small lake and a car completed the scene.
    The next step, was to convert all Autodesk materials to an Arch&Design set, in order to access the fine tune settings in the materials properties, so we can archive a full speed Max Mentalray version. The next logic step, was the Iray version (given the same set of Arch&Design materials). Klaus also remade from scratch the materials and the lights for another 2 rendering video engines: Vray and Octane (max plugin).

    We are not 3dmax experts, but this is what we succeed to achieve so we can test our hardware.

    In the archive(see the link), you will find 4 versions of this scene each for every engine: Vray, Mray, Iray, Octane.
    In order to take a bechmark follow this steps:
    1-open the scene version you want to test it.
    2-you will receive a Missing Dlls message (a 3DMax bug when using scenes that have fbx links inside)
    3-press Cancel
    4-open the scene again (it will work this time)
    5-press Render Button (or Shift+Q)
    6-when the rendered is finished, take note about the time it took to render to complete (the info is in the lower left part of the 3dmax main application window, in the status bar)
    7-give us info about that rendering time, about your hardware specifications and about the hardware components that were used by the rendering engine (last info very important for Iray&Octane)
    In order to keep the info organized, copy paste the below example and change the text accordenly.

    Example:
    --------------------------------------------------------------------------------------------

    Rendering Engine: Iray v2.0
    Rendering using: CPU+GPU1+GPU2
    Rendering Time: 0:12:03

    OS/System RAM amount: Windows 7 64 bits Ultimate/16GB
    Software: 3d MaxDesign 2012 SP2
    CPU: Intel I7 3930K (stock frequency)
    GPU1/Video RAM amount: Nvidia GTX560TI/2GB (stock frequency)
    GPU1 Model (optional): Palit GTX560TI 2048M
    GPU2/Video RAM amount: Nvidia GTX560TI/2GB (stock frequency)
    GPU2 Model (optional): Palit GTX560TI 2048M

    --------------------------------------------------------------------------------------------

    If you post more than one benchmark (in the same post) you don`t need to have the hardware spec listed again so use only the next short info:

    --------------------------------------------------------------------------------------------
    Rendering Engine: Iray v2.0
    Rendering using: CPU+GPU1+GPU2
    Rendering Time: 0:12:03
    --------------------------------------------------------------------------------------------

    Additional explanations:

    As you know:
    Mray and Vray are CPU benchmarks (biased rendering - simplified algorithms)
    Octane is GPU benchmark. (unbiased rendering - physically corect algorithms)
    Iray is GPU/CPU benchmark. (unbiased rendering- physically corect algorithms)

    The quality for the Mray and Vray versions has been set in such way so a normal machine to render somewhere under an hour.
    For the Iray engine there is a limit of 500 itrerations. For the Octane version the limit is set to 1500 passes. So the final image will be a little noisy. The difference between number of iterations (500) from Iray and the number of passes(1500) from Octane is related to the fact that we think (based on our observations), that an Iray iteration is doing more calculations(in terms of quality) than an Octane Pass. Also the limit number for bouth engines had to be on a medium video card under an hour.


    Good luck with testing your hardware, and let us know the results !!

    Cheers





    ___________________
    PS
    If you take the benckmark for the Iray/Octane engines, read the info below. If not, don`t waste your time reading.

    Minimal video card memory amount requirement : 1GB (The scene has been tested to fit in 1GB video memory (nvidia fermi ), installed on a Windows 7 64bit, with windows 7 basic theme applied, having 1 monitor configuration with a maximum resolution of 2560x1600).

    In Iray if you have Iray 2.0 (available only if you install 3dMax 2012 extension pack from subscription center), we are interested in the first place for only GPU benchmark, and secondary, if you have time, GPU+CPU bachmark. Only CPU benchmark is not need it.

    In Iray and Octane if you have multi video cards configurations, it will be nice to take benchmarks with different combinations of you cards: GPU1 banchmark, GPU1+GP2 benchmark, GPU1+GP2+GPU3 benchmark, etc.

    How to detect if the GPU engine is using your card (it is very possible to not use it):
    Iray is a tricky engine. In Iray 2.0 version (available only if you install 3dMax 2012 extension pack from subscription center) you could choose what hardware components to allocate for rendering. Like how many video cards and how many cores from your CPU. So, if your video card has just 1GB memory or a little more, and your video memory is already filled with other programs staff, and if this max scene don`t enter in that video memory that is still free, than, iray, will not use your video card, only the CPU (even if you selected to use that video card in the Iray settings>Hardware resources). So in order to be sure that all your selected video cards are used, it is a good idea to install a small free utility like GPUtemp (GPU Temp) to see if the temperature of yours GPU are rising once the image of the model appears. If one of your video cards don`t have that rising temperature , than that video card is not used.

    In Octane is almost the same thing, the only difference is that Octan don`t make use of CPU. So you have to take care so the scene can fit in at least one video card memory, in order to take the benchmark (it will use only those cards that can receive the entire scene in theirs memory).

    It is very important to identify what components your Iray/Octane engine succeeded to use, so the info about your render time can have a logical meaning.

    Some tips to be sure that the scene succeed to transfer in your limited video card memory (usually important for 1GB or 1.25GB cards, but maybe also for bigger memory amounts if you have a lot of windows opened).
    -close all unnecessary programs
    -set one monitor configuration
    -if your OS is Win7 choose the Win7 basic theme.
    -If your OS has a custom background like a high resolution digital photo, remove it temporary with a simple color.

    www.revitforum.org/rfo_renderbenchmark.zip
    Attached Files
    Last edited by gaby424; February 25, 2012, 07:19 PM.

    #2
    reserved space for listing future results...

    Comment


      #3
      To get this rolling, here's my benchmark results. Hope everyone with a license to Max (or just a trial version) will participate in this! And thanks Gaby, for the hard work you've put into it.

      Rendering Engine: Mray
      Rendering using: CPU
      Rendering Time: 0:19:06

      Rendering Engine: Vray
      Rendering using: CPU
      Rendering Time: 0:23:05:

      Rendering Engine: Iray v2.0
      Rendering using: CPU
      Rendering Time: 1:17:58

      Rendering Engine: Iray v2.0
      Rendering using: GPU2
      Rendering Time: 0:21:46

      Rendering Engine: Iray v2.0
      Rendering using: GPU1+GPU2
      Rendering Time: 0:11:04

      Rendering Engine: Iray v2.0
      Rendering using: CPU+GPU1+GPU2
      Rendering Time: 0:10:15

      Rendering Engine: Octane MAx
      Rendering using: GPU1+GPU2
      Rendering Time: 0:09:20

      OS/System RAM amount: Windows 7 64 bits Pro/16GB
      Software: 3d MaxDesign 2012 SP2
      CPU: AMD 1090T 6x3,2GHz (Overclocked to 3,9GHz)
      GPU1/Video RAM amount: Nvidia GTX570 1,25GB (stock frequency)
      GPU2/Video RAM amount: Nvidia GTX570 1,25GB (stock frequency)
      Last edited by Munkholm; February 6, 2012, 08:06 PM.
      Klaus Munkholm
      "Do. Or do not. There is no try."

      Comment


        #4
        Rendering Engine: Mray
        Rendering using: CPU
        Rendering Time: 0:11:43

        Rendering Engine: Iray 2.0
        Rendering using: CPU
        Rendering Time: 0:30:16

        Rendering Engine: Iray 2.0
        Rendering using: GPU1
        Rendering Time: 0:29:34

        Rendering Engine: Iray 2.0
        Rendering using: CPU+GPU1
        Rendering Time: 0:19:58

        OS/System RAM amount: Windows 7 64 bits Ultimate/16GB
        Software: 3d MaxDesign 2012 SP2
        CPU: Intel I7 3930K (overclocked @ 4.3ghz)
        GPU1/Video RAM amount: Nvidia GTX560TI/2GB (stock frequency)
        GPU1 Model (optional): Palit GTX560TI 2048M

        Intersesting is that CPU is almost equal with GPU in my case(in Iray).
        Also, Klaus it seems that 2 GPU is twice faster than 1 GPU in your iray test, wich is a nice scaling.
        Last edited by gaby424; February 7, 2012, 07:41 AM.

        Comment


          #5
          Originally posted by gaby424 View Post
          Intersesting is that CPU is almost equal with GPU in my case(in Iray).
          Not surprised at all... Your CPU is a monster running 12 cores OC'ed to 4,3GHz and your GPU have relatively few CUDA cores.
          Klaus Munkholm
          "Do. Or do not. There is no try."

          Comment

          Related Topics

          Collapse

          Working...
          X