ScienceMark 2.0 is a benchmark based on real molecular dynamics problems. However, unfortunately, it only creates two computational threads, which has direct effect on the results. Dual-core and quad-core Intel processors perform almost equally fast here.
Following numerous requests from our readers we resumed processor testing in popular math1ematical suites, namely MATLAB and Mathematica. The latest versions of these applications are optimized for multi-threaded work.
The results of MATLAB optimization appeared quite unexpected. Although some operations (such as linear algebra, for instance) start working much faster with multi-threaded support activated, other calculations slow down. This slowing obviously affects the integral result you see on the charts for CPUs with four computational cores.
As for the Mathematica optimizations for multi-core processors, we can declare that it was pretty successful.