The benchmark results did not give us any major surprises when it comes to performance. Applications that have been optimized for multi-core processors by thread optimization scale beautifully, from one to four processing cores. Looking at the benchmarks is also a nice way of making clear which applications have been thread-optimized - and reveal those that weren't.
We don't have an explanation for the issues with the synthetic benchmark suite Sandra by SiSoftware. Its low-level CPU arithmetic and multimedia benchmarks, as well as the memory benchmarks, refused to work with three cores. It remains to be seen whether this is an issue with our test system - on which we simply disabled individual cores in the BIOS - or with three cores in general.
The same applies to power consumption results. Although individual cores were switched off, the overall system power consumption remained almost the same - this is an indicator that the motherboard BIOS cannot physically shut down cores. Hence we have to wait for final version of AMD's Phenom triple core to make a final conclusion.
Since performance does not vary a lot between three and four cores with most applications, we expect the triple core to be a nice alternative for energy-efficient solutions that still require a lot of computing power. It will be interesting to see if the triple core could actually offer better energy efficiency than the quads.