Search results

Found this at Ars while browsing around today. It appears that PCMark05 uses different optimizations for the memory benchmark based on a processor's CPUID. The article is linked below. Just thought that this was another example of the problems with canned benchmarks, as opposed to real-world...