Customers often ask us what hardware they should consider purchasing, or which hardware is faster than another. This is a very difficult question to answer; since the answer is completely dependent on what applications you are going to use your hardware for. In our experience, one application may run very well on a machine, while another will not run as well. It is seldom possible to base machine performance only on one application - the user must consider all uses of the machine. Because of this, we have decided to give you a EnSight command file which can be run on your workstation and compare its performance against other hardware.

Show me the numbers!

EnSight performance testing was started several years ago. A great number of performance benchmarks have been completed over the time. However, the previous two benchmarks are not longer large enough to adequately test new hardware. Accordingly, a new benchmark has been created - significantly larger than the previous tests.

CEI does not endorse or recommend any hardware vendor above another. As an independent software company we supply software to run across multiple hardware platforms. We can, however, report to you some benchmark numbers on how well the graphics hardware of various workstations perform, and also submit to you a EnSight version 8 command file which can be used for benchmarking graphics hardware.

The command file in this directory, performance.enc (win_performance.enc for Windows) can be used to perform five tests that will stress the graphics system as well as the CPU and memory systems.

All five tests operate on a geometry containing three parts containing a total of 6324 quads and 128,534 triangles. These parts are then duplicated for a total of six instances. Thus the total number of polygons in the test is 37,944 quads and 771,204 triangles. All ts are run using a display area measuring 600x500 pixels. All polygons are randomly oriented (i.e., no "stripping" is done).

The first test is a line drawing test that rotates the scene 360 degrees in 12-degree increments for 30 refreshes of the screen. The total number of lines drawn during the test is (37,944*4 + 771,204*3)*30 = 73,961,640 lines. Each part has a single color.

The second test is a shaded test that rotates the scene 360 degrees in 12-degree increments for 30 refreshes of the screen. The total number of polygons drawn during the test is 24,274,440. Each part has a single color.

The third test is a repeat of the second test, but here the parts are colored on a per vertex basis.

The fourth test is a repeat of the third test, except it is run in immediate mode (as opposed to display list mode for the previous tests). EnSight uses immediate mode for flipbook animations, hidden line display, and all detached (VR) displays.

The fifth test is a repeat of the fourth test, but here the two large isosurface parts are transparent and the rotation uses 72 degree increments for a total of 5 refreshes. This test stresses not only the graphics subsystem, but the cpu/memory as well since the polygons have to be sorted for each refresh of the screen.

To run the performance test, start EnSight from this directory as follows:

ensight8 -no_prefs -glconfig -p performance.enc

Use win_performance.enc for windows. Once the tests are completed, you will see some timing information in the shell window where you started EnSight that looks something like:

[front animate on] [front animate off] time in seconds = 29.705112

repeated for each of the five tests. Of interest is the "time in seconds".

A composite score can be computed for each platform by using the frame rate and weighing the importance of each test as follows (the list is in performance order):

    * Test 1: 25%
    * Test 2: 20%
    * Test 3: 20%
    * Test 4: 20%
    * Test 5: 15%

So the composite score, C, using test1 (T1) through test 5 (T5) is found as follows:

C = 0.25*(30.0/T1) + 0.2*(30/T2) + 0.2*(30/T3) + 0.2*(30/T4) + 0.15*(5/T5)

It is important to note that these tests only measure the time required to complete the animations. Image quality is more subjective and may be better on some systems than others. In particular, on some systems multi- sampling is enabled by default (either by EnSight or through the graphics driver). The resulting image quality is much better, but there can be a cost in performance. If you are duplicating these tests, you might try the following command-line options, which can have an effect on the rendering performance:

• -multi_sampling     
• -no_multi_sampling     

In cases where EnSight controls the multisampling, these options will change the default setting. In cases where the graphics driver controls multisampling, check your documentation to find the environment variable, configuration file, or command that controls the setting.

• -ogl
  
• -gl

In wireframe mode (test1) EnSight can either draw the wireframe models as lines (-ogl) or as hollow polygons (-gl).

• -norm_per_vert     
• -norm_per_poly     

When drawing flat-shaded polygons (test2-test5), EnSight can send one normal per polygon or one normal per vertex. While the results are the same, some graphics drivers handle one case better than the other.

If you find that the default EnSight mode is not optimal on your machine, or you want to submit performance results on a system not shown here, please send your results and system information to This email address is being protected from spam bots, you need Javascript enabled to view it . Please include both the timing information that is printed by EnSight as well as the output, which results from the -glconfig option:

GL configuration:

vendor string: NVIDIA Corporation

renderer string: Quadro2 Pro/AGP/SSE

stencil buffer: enabled

depth buffer: enabled

wireframe mode: hollow polys

display list: OpenGL

normal per vertex: off

Below are the scores for the individual tests:

* The dark green bands indicate Windows machines.  All others machines are in yellow.