Cinebench R10

Website: Cinebench R10

Maxon Cinebench is based on Maxon's popular animation software, Cinema 4D, which is used extensively by studios and production houses worldwide for 3D content creation. We've used the built-in CPU benchmark, which uses a 3D scene file to render a photo-realistic image of a concept bike. The scene makes use of various CPU-intensive features such as reflection, ambient occlusion, area lights and procedural shaders.

Euler3D

Website: Euler3D

We also ran the Euler3D HPC benchmark from the CASE lab at Oklahoma State University. This benchmark analyses airflow over a section of the wing of an aircraft in a virtual wind tunnel travelling at Mach 0.5 (half the speed of sound) using CFD (computational fluid dynamic) calculations. Euler3D presents its results expressed as a CFD cycle frequency in hertz, so a higher number is better.

FlamMap

Website: FlamMap

Another HPC application is fire behaviour mapping and analysis. FlamMap is used by several federal government organisations in the USA, and uses a terrain map combined with meteorological data to compute fire behaviour (likelihood, location, intensity and movement) over time.


In the Cinebench multi-threaded CPU rendering test the Opteron 2435s finally showed their mettle, scoring 25,595, a substantial improvement over the 18,563 the older Opteron 2382s scored. However, the Xeon W5580s were once again much faster, scoring 29,980.

Euler3D expresses its results as a CFD cycle frequency in hertz, so a higher number is better. Once again the four additional cores and higher memory bandwidth of the dual Opteron 2435 system gave it a significant lead over the dual Opteron 2382 system, with a score of 4.005Hz versus 2.780Hz. However, the Xeon W5580s are a real beast at CFD type calculations, achieving a monstrous score of 6.897Hz.

The six-core Opteron 2435s also proved much faster than the Opteron 2382s in FlamMap. They completed this test in 177 seconds, versus a pathetic 310 seconds by the Opteron 2382s. The Xeon W5580s were once again faster though, as this system finished the test in 130 seconds.