In theory, dual-channel gives you a combination of twice the bandwidth and half the cache line fill latency. In practice, I do not remember seeing figures about that in a long time but I would expect that to make a 10-15% difference in typical benchmarks, more for graphics-intensive benchmarks on IGP or memory-intensive apps.