Benchmark Results: Scientific Tests
One of the other major uses for workstation-class PCs is in the scientific, CAD, and engineering markets. We’re working on tests to fill in this part better, but to start with we’ve used a standard Euler3D engineering benchmark.
This Euler3D benchmark was developed by the Computational AeroServoElasticity (CASE) Lab at Oklahoma State University’s Department of Mechanical and Aerospace Engineering. The test simulates iterations of a computational Fluid Dynamics (CFD) airflow problem over a NACA 445.6 aeroelastic test wing at Mach 0.5 (half the speed of sound).
The test reports two figures: the number of iterations the test completes per second (reported in Hertz) and the amount of time necessary to complete the benchmark (in seconds).
The Euler3D benchmark only gains about 9.5% from Hyper-Threading, but is fully threaded. This is indicative of another possible bottleneck that prevents us from realizing a more substantial gain--or perhaps Hyper-Threading simply isn't able to stretch its feet here due to more complete utilization of each physical core by each thread of the benchmark.