Comparing dual and single CPU systems on the basis of a benchmark application that only measures the raw increase in performance is rather pointless as it is far from a real-world situation. Therefore we’ve chosen to use Bapco’s SYSmark2002 to measure overall system performance and as a means of creating a somewhat ‘typical’ user environment, this to determine how much performance gain a typical user would have from a dual CPU system.
The results from the SYSmark2002 benchmark clearly show that the new dual Xeon turns in the highest score, followed by the AMD and with the two single CPU systems from Intel and AMD closing the ranks. The benchmark score from the dual Intel Xeon is pretty convincing as it is over 19% (220/184) faster than the dual AMD Athlon. However the differences between the single CPU systems are less convincing, there’s only a 2.9% (178/173) gap between them, not enough to favor one over the other as this will go unnoticed during every day use. It is interesting though to look at the differences between single and dual CPUs systems both for AMD and Intel. The performance increase for the dual AMD Athlon over the single AMD Athlon is a mere 6.4% (184/173), whereas the dual Intel Xeon is an impressive 24% (220/178) faster than the single Pentium 4.
So judging from the SYSmark2002 benchmark we can safely say that a dual AMD Athlon doesn’t offer much of a performance increase over a single AMD Athlon, 6.4% more performance isn’t something to get worked up about, especially when looking at the investment we need to make in order to build such a dual AMD system. For the amount of money spent on the 2nd CPU, the motherboard and all other extras, we could easily have built a single AMD Athlon system with higher performance parts that would run circles around the dual. Unfortunately the same holds true for the dual Intel Xeon, as although the performance increase makes the upgrade to a dual Intel Xeon worthwhile, we’re quick to realize that the high price of such a system in fact kills the price vs. performance ratio, it simply is too expensive in comparison to a single CPU system.
The other benchmark we’ll use to quick test game and 3D application performance is 3DMark2001 Second Edition from MadOnion. 3DMark2001 is naturally meant for testing graphics performance and although it is not SMP aware that’s exactly what we’ll be using it for. We’ve used one of the fastest graphics cards currently available, an Asus GeForce4 Ti4600, to make sure that any variations in benchmark scores aren’t due to the graphics card but can solely be attributed to the system used during testing.
Again the dual Intel Xeon turns in the best results, whereas the single AMD Athlon is the 2nd fastest, both the dual AMD Athlon and the Intel Pentium 4 are somewhat slower. This benchmark clearly shows that a single AMD Athlon is faster than a dual AMD Athlon when running 3D games and DirectX applications. The single Pentium 4 also lags behind somewhat and is bested by the single AMD Athlon and the dual AMD Athlon.