Adobe software developers continue optimizing their applications for multi-core architectures. However, in reality it turns out that the most “heavy-duty” filters still cannot really use the advantages of multi-core CPUs. Therefore, dual-core Intel processors with higher clock frequency and individual core efficiency defeat triple- and quad-core competitor solutions.
The things do not look too good in Mathematica 7, too. Although this application learned to split the calculations into parallel thread within a single kernel, dual-core Intel processors work better in this mathematical suite. However, you shouldn’t forget that multi-core processor architecture may come in very handy when there are several kernels launched at the same time.
The WinRAR results correlate in very interesting way. The performance of this archiving utility depends on the number of CPU cores, but is also very sensitive to the cache-memory subsystem configuration. Therefore, dual-core Core 2 Duo E8400 processor with high-capacity and high-speed shared 6MB L2 cache can successfully compete not only with triple-core Phenom II X3 720, but also with a quad-core Phenom II X4 810. As we know, AMD processors yield quite seriously to dual-core Intel solutions from the Wolfdale family in terms of cache-memory subsystem efficiency. At the same time Core 2 Duo E7000 processors with a cut-down L2 cache and slower bus lose to the identically priced AMD solutions.
In Excel the third core of Phenom II X3 processors becomes a sufficient compensation for the microarchitectural advantages of Core over K10, but no more than that.
However, AMD’s triple-core concept wins a real victory in the applications with well-paralleled load that do not depend on the cache-memory subsystem performance that dramatically. And the examples are right here: chess benchmark or Folding@Home are excellent illustrations to this statement.