iXBT Labs - Computer Hardware in Detail






Open source method of measuring rendering performance in 3ds max 5.1 SP1

First of all, to prove that the test suite is not that artificial, I must say that it has a certain plot. 

So, when a guy wants to build (or disassemble) anything he goes to a garage. It's the first reason why a garage is a polygon for our tests. The second reason is that a garage is an excellent dump of all possible things. So, we have a shabby garage with an old auto (to realize as many effects as possible :). Some stuff, including a number of tools each of which fully uses one of the renderer's features, hangs on the walls. The most important tool is a workbench where various experiments are carried out. I.e. every effect can be examined on several different objects which, in their turn, could use one of them. 

Scenes for Brazil r/s

1. Brazil_Caustic

This scene tests the CPU speed by rendering the Caustic effect with the SplurtterFish Brazil renderer. The object is a monument-souvenir made of glass with the imprinted memorable sign and a wooden desk. For the desk the power of photons was increased up to 1.7 (Power=1.7), all the other settings are default.

2. Brazil_DOF

This scene tests the rendering rate for the Depth Of Field effect. The scene is simple: it uses three objects located at different distances from the camera, with the central one being in focus. All the settings are default except: 

  • Initial Rate: 6
  • f-Stop: 8,0
  • Focal Dist: 70
  • Bokeh: Flat Disc

3. Brazil_GI

GI - Global Illumination. It actually uses the Brazil_Final scene (see below) but the final scene doesn't contain any textures so that we can focus on the global illumination scene. All the settings are default.

4. Brazil_Final

The final scene displays a realistic version, i.e. the product which can be created by a designer when he works on a real task. This scene doesn't use such trendy stuff like caustic effects but uses the effects most popular among 3D designers: Global Illumination, Antialiasing, Raytrace, Volume Light etc. Settings:

  • Global Illumination
    • Shade Rate
      • Min: -4
      • Max: -4

    • Sampling Control
      • View Rate: 6
      • Sec Rate: 2

  • Antialiasing
    • Min Samples: 1
    • Max Samples: 2

  • Volume Light
    • Density: 1
    • Sample Volume: 60

  • Attenuation
    • End: 85
    • Res: 1024x768

5. Brazil_Preview

This subtest tests operations frequently fulfilled by designers - they reduce the resolution and quality settings in order to increase the rendering speed and visually estimate the outcome. The preview performance is as important as the final rendering because it has to be carried out a dozen of times (the final rendering ideally takes place only once). Brazil_Final scene with the following settings:

  • Resolution reduced to 800x600
  • Anti-aliasing disabled
  • Luma Server has the following parameters changed:
    • View Rate: 4
    • Sec Rate: 1

  • Sample Volume (in the volume light settings) reduced to 20

Scenes for Final Render Stage-1

1. FR_Caustic

The scene is identical to the one for Brazil renderer, though in this case it uses the Final Render. The light source settings are modified: 

  • Multiplier: 2
  • Shadow: fr-Area Shadow
  • Final Render Photon
    • Energy: 20
    • Light Rays: 200000

2. FR_GI

The scene looks familiar - we saw a similar picture in the scene of testing the global illumination speed with Brazil. This test uses a different engine (Final Render / fr-image). The GI settings are default, only the AA settings are changed: 

  • Sampler: Adaptive
  • Min Samples: 1/4
  • Max Samples: 4
  • Filter: Catmull-rom

3. FR_Displace

This scene allows estimating the testbed performance when rendering the Micro Triangle Displacement effect. The key objects are a logo on the support and an ordinary ball lying on the workbench used if the Displace effect is not that well seen with the logo. The following default settings were modified:

  • Subdivision level: 50
  • Edge Length: 3,0


One more single-effect scene, Sub-Surface Scattering rate is measured. The parameters modified:

  • Depth: 5
  • Absorption: 15
  • Anisotropy: 10
  • Multiplier: 30

5. FR_Volume_Light

fR-Volume Light. The test is primitive: a volume light source and holes. The following default settings were modified:

  • Density: 2,0
  • Shadows
    • Raytracing smp: 20

  • Attenuation Color: on
  • Samples: 10

6. FR_Volume_Caustic

This test estimates the rendering rate of the volume caustic (the surface one is also enabled). A ball or a magnifier suit best of all for this test. As a result, the designer made a kind of a ball-shaped magnifier. Settings:

  • Volume light:
    • Step Size: 5
    • Accuracy: 60
    • Multiplier: 5

  • Light source:
    • Multiplier: 2
    • Light Rays: 20000

7. FR_Final

The idea is similar to Brazil_Final, except the renderer. Settings:

  • GI - without any changes
  • Volume light:
    • Shadows
      • Raytracing smp: 60

    • Attenuation Color: on
    • Samples: 10
    • Density: 1

  • 4 light sources

8. FR_Preview

Again a draft version of the final scene. The following settings are changed (reducing the rendering quality):

  • AA disabled
  • Resolution reduced to 800x600
  • GI settings:
    • Bounces: 1
    • Size Ratio: 1/8
    • Balance: 50%

Scenes for Scanline (3ds max 5 renderer)

1. Scanline_Volume_Light

The idea is identical to FR_Volume_Light, only the renderer is taken from 3ds max. Settings:

  • Volume:
    • Density: 1
    • Sample Volume: 60
    • Attenuation Color: on

  • Attenuation:
    • End: 85

2. Scanline_Light_Tracer

Simple light provided by the 3ds max renderer. Settings - Ray/Sample: 300.

3. Scanline_Radiosity

This is a more advanced version of Global Illumination supported by 3ds max renderer. Note that a light map must be applied before rendering with Scanline renderer. It takes the most part of time, that is why the given test is the only one where we use a stopwatch. 

...first we select a required item in the menu...

...press "Start" and note the time...

...get the calculated light map with the given parameters...

...and then press F9 to get the final results.

The result is the total time of execution of Advanced Lighting --> Radiosity and operation of the renderer. Settings:

  • Process
    • Refine Iterations All Object: 3

  • Global Subdivision Settings
    • Mesh Size: 15

4. Scanline_Final

The final scene is identical to the two previous ones - the developer says that this is a "parody of GI from Scanline in the form of Light Tracer". There are also several materials with the Raytrace effect. Plus, AA by the same Scanline renderer. Parameters:

  • GI: Light Tracer:
    • Ray/Sample: 300
    • Antialiasing: Catmull-rom

  • Volume light:
    • Volume Density: 1
    • Sample Volume: 60
    • Attenuation Color: on

  • Attenuation:
    • End: 85

5. Scanline_Preview

The standard preview algorithm of accelerated rendering with some variations:

  • Global Illumination disabled
  • Sample Volume reduced from 60 to 2
  • AA disabled
  • Resolution reduced to 800x600


AA parameters on all scenes for Brazil except Preview:

  • Min Samples: 1
  • Max Samples: 2
  • Filter: Mitchell-Netravali

AA parameters on all scenes for Final Render except Preview:

  • Sampler: Adaptive
  • Min Samples: 1/4
  • Max Samples: 4
  • Filter: Catmull-rom

AA parameters on all scenes for Scanline except Preview:

  • Filter: Catmull-rom


The technique developed by the guys from the iXBT.com CPU section and 3D designer Igor (mazai) Emelianenko fully tests performance of modern computer systems based on the x86 architecture at the final scene rendering in one of the most popular 3D design suites - 3ds max.

In contrast to most other techniques where the final rendering in 3ds max is considered as one of subtests of the general-purpose test technique this one uses 18 scenes and three popular renderer for 3ds max - one integrated and two external,  instead of just one scene for one renderer. 

So, in spite of the complex and time-taking tests this technique allows to deeply study the issue of performance of computer systems in 3ds max. On the other hand, <...>_Final scenes are very close to realistic models, and this fact lets us shorten the test technique to 3 tests to make it a part of a more universal technique.

Notes, additions, files, and materials based on the test technique:

  1. Test scenes for 3ds max
  2. Textures for the test scenes
  3. Benchmarking: top x86 CPUs of all architectures, Part 1.

Stanislav Garmatyuk (nawhi@ixbt.com

Write a comment below. No registration needed!

Article navigation:

blog comments powered by Disqus

  Most Popular Reviews More    RSS  

AMD Phenom II X4 955, Phenom II X4 960T, Phenom II X6 1075T, and Intel Pentium G2120, Core i3-3220, Core i5-3330 Processors

Comparing old, cheap solutions from AMD with new, budget offerings from Intel.
February 1, 2013 · Processor Roundups

Inno3D GeForce GTX 670 iChill, Inno3D GeForce GTX 660 Ti Graphics Cards

A couple of mid-range adapters with original cooling systems.
January 30, 2013 · Video cards: NVIDIA GPUs

Creative Sound Blaster X-Fi Surround 5.1

An external X-Fi solution in tests.
September 9, 2008 · Sound Cards

AMD FX-8350 Processor

The first worthwhile Piledriver CPU.
September 11, 2012 · Processors: AMD

Consumed Power, Energy Consumption: Ivy Bridge vs. Sandy Bridge

Trying out the new method.
September 18, 2012 · Processors: Intel
  Latest Reviews More    RSS  

i3DSpeed, September 2013

Retested all graphics cards with the new drivers.
Oct 18, 2013 · 3Digests

i3DSpeed, August 2013

Added new benchmarks: BioShock Infinite and Metro: Last Light.
Sep 06, 2013 · 3Digests

i3DSpeed, July 2013

Added the test results of NVIDIA GeForce GTX 760 and AMD Radeon HD 7730.
Aug 05, 2013 · 3Digests

Gainward GeForce GTX 650 Ti BOOST 2GB Golden Sample Graphics Card

An excellent hybrid of GeForce GTX 650 Ti and GeForce GTX 660.
Jun 24, 2013 · Video cards: NVIDIA GPUs

i3DSpeed, May 2013

Added the test results of NVIDIA GeForce GTX 770/780.
Jun 03, 2013 · 3Digests
  Latest News More    RSS  

Platform  ·  Video  ·  Multimedia  ·  Mobile  ·  Other  ||  About us & Privacy policy  ·  Twitter  ·  Facebook

Copyright © Byrds Research & Publishing, Ltd., 1997–2011. All rights reserved.