The OCUS Benchmark of Pro/E Builds

Results of running the OCUS Benchmark v3.2 with different builds of Pro/E.

The v3.2 Benchmark has trail files for R2000i up to R2001. I originally released some Wildfire trailfiles, but Michael Campbell of PTC pointed me to a number of errors in the Wildfire trailfiles.

The most important error was that only the Wildfire trailfiles created a pattern of 100 datum planes (a feature from a previous benchmark version) in the generic part which resulted in 10.000 (!) extra datum planes in the assembly.

Here are the remarks I received from Michael Campbell from PTC.

PTC and I decided to rerun the test with specially created trailfiles for Wildfire. These trailfiles will not be released, since they are too complex to run.

In order to get good comparible results the benchmark was split up in a Wildfire trail file for all task which had to be done with the Wildfire GUI and an R2000i trailfile specifically for the pan, zoom and spin tests. Both trailfiles were run with Wildfire build 2003051.

So, I finally ran the R2000i, R2000i2, R2001 and Wildfire version trail files of the v3.2 benchmark on one computer under more or less the same conditions.

The tests were performed on a:

Dell PWS650 P4 Xeon 2.40GHz;
nVidia Quadro4 900 XGL (128Mb) (driver 6.13.10.4104);
1.5Gb RAM;
Maxtor 36Gb SCSI disk;
WinXP SP1 (build 2600);

The results show how the OCUS Benchmark performance of Pro/E decreased by 12% compared to R2001.
The following tasks were all 50% or more slower compared to R2001:

• hiddenline redraws with FastHLR (GR)
• shaded mouse spins (GR)
• shaded redraws (GR)
• wireframe mouse zooms (GR)
• Explode translations (GR)
• Screen Updates (GR)
• Perspective Views (GR)

More on these differences in the remarks by Michael Campbell

• Shade Calculations (CP)
• Menu Redraws (GU)