Get best of the week

Testing video cards to work with KOMPAS-3D

Continuing the theme of iron for CAD. In the last article, we tested processors while working in KOMPAS-3D, today we will focus on video cards. In touch again Stanislav Ermokhin, head of the department of technical support and training of the regional center ASCON-Volga .



We noted earlier that the speed of the KOMPAS-3D system depends on the characteristics of individual components. Our internal test, created at KOMPAS Lab, allows you to compare different hardware on the same model with the same sequence of actions.

Our test suite contains a number of commands and actions that most users have to perform in 3D modeling of parts and assemblies. The implementation of the first two stages of these tests falls entirely on the central processor, namely:

  • discovery of an experimental model;
  • rebuilding the model to update the changes made and the actual calculation of mass-center characteristics.

But there is a third stage, during which the model constantly moves or rotates in all possible options for displaying the model (frame, grayscale, etc.), and here at this stage the main load falls on the video card.

A little more about the benchmark from the developers:
Turchin Roman , Head of Testing Department KOMPAS-3D
Aleksey Nikonov , Analyst Engineer KOMPAS-3D:

Testing the video subsystem is carried out by measuring FPS when rotating the model in five rendering modes:

  • halftone with frame
  • halftone
  • frame
  • frame without invisible lines
  • frame with thin invisible lines

We consider the grayscale mode with the frame to be the main one when working in 3D. The results in this mode are multiplied by the weight coefficient. Accordingly, this mode affects the overall results the most.

So, the test sequence is similar to the previous testing, the model too. Before testing video cards, we note that the triangulation of the model consists of 2,300,000 (two million 300 thousand) polygons. The number of edges and faces we have not yet calculated - each time we get lost in the fourth million :).

The rating scale will differ from the scale of the previous processor rating test. The rating of the video cards of our automatic test must be considered through the following ratings:

  • 25 or less points - a frankly weak result for a video card;
  • up to 75 points - average result;
  • over 75 points - a good result for a video card that allows you to solve problems of your level.


And here are some participants in the tests

“And with what settings / parameters of the system are tests being carried out?” - a logical question arises.

To track all the features of the latest versions of KOMPAS-3D, we have the following settings:

  • auto-detection of the rendering option - for all modern video cards (those newer than 2012), the “Improved” option will be automatically selected;
  • smoothing quality "Medium" (8x MSAA);
  • simplified display mode is disabled - we will hurt KOMPAS. In the same mode, lines of essays are drawn, creating an additional load;
  • resolution FullHD monitor - 1920x1080.

After the measurements, the summary diagram of the test results is as follows. Summary chart. (The picture is clickable) In the general table, the eyes run wide. Let's specify the results for manufacturers and families of graphic processors for proposals that are relevant on the market.






Nvidia Graphics Cards


Pascal Graphics Cards



(upd: graph updated)

Turing Graphics Cards




AMD Graphics Cards



* AMD Radeon RX 5000 next-generation cards highlighted in orange

What conclusions can be drawn from the presented graph?


  1. , , : . .
    , .
    AMD Radeon(TM) Vega 11 Graphics “”. , , — .
  2. AMD Radeon RX 560 NVIDIA GeForce GTX 1030. , . .
  3. — . NVIDIA GeForce GTX 1050, NVIDIA GeForce GTX 1050Ti, AMD RX 570 Radeon RX 580.
    ( ) (2000-5000).
  4. NVIDIA GeForce GTX 960 NVIDIA GeForce GTX 780. . NVIDIA GeForce GTX 780 , OpenGL 4.5, , .
  5. AMD RX 5500XT , .

    NVIDIA GeForce GTX 1060 Turing: GTX 1660/1660Ti / .

    AMD Nvidia , . AMD AMD Radeon RX 5700. , , AMD Radeon RX 5700 XT (, ) , NVIDIA GeForce RTX 2060 SUPER.

    , , , .

  6. - , NVIDIA GeForce RTX 2070 NVIDIA GeForce RTX 2060 SUPER. , . 2070, , ( 8+6 pin, 8 pin, .. ). 2060 Super, . .
  7. , , NVIDIA GeForce RTX 2080 SUPER. . , , AMD RX 5900 XT . .

    , — AMD RX 5500XT NVIDIA GeForce RTX 2080 SUPER, . , : , , OpenGL 4.5, 4 , (Memory Bandwidth ) — 140 / .

    4K , , -3D . , .



    At 4K resolution, video cards, although they lost from 10% to 20% of their performance, still allow you to work comfortably in KOMPAS-3D. When choosing a video card for this resolution, we recommend giving preference to models with a large video buffer size of 6-8 GB.

General conclusions:

  1. Only gaming versions of video cards participated in the test.
  2. Integrated graphics cards are not suitable for full-fledged work.
  3. Modern gaming graphics cards are suitable for use in professional design systems such as KOMPAS-3D.
  4. Using budget solutions (“plugs”) is justified if you work only with drawings and the simplest assemblies. With the complexity of the tasks performed, a more serious video card will be required.

In the future we plan to continue testing the components - we will write about the results in the following articles.

ASCON thanks AMD and Konstantin Voron personally for their help in preparing the material.

More articles:


All Articles