We wanted to see whether the increased L2 cache latency of the 65 nm Brisbane core made a difference, so we focused our benchmarking on memory and L2-sensitive applications and synthetic tests. Clearly, there are some differences, but they're minor.