IIRC the AMD memory controller was bandwidth starved and showed slightly different results.
The trouble with the benchmarks you list is that as memory speed climbs, the latency climbs too. Given the same latency (as in my case) the higher speed brings additional benefits.
Memory is about access, so a tradeoff between speed and latency is like using a very small towel to cover yourself. something begins to show at either end. You can see the effect of the lower latency kit even in the benchmarks provided.
'Much' difference is subjective and depends on the benchmark. In real life memory is never really used the way a benchmark indicates it to. It is totally slaved to the operating system, and what you need for real objective memory evaluation is a AT-style (of old) multitasking test, which is possibly the closest you can get to real life.
It has been a much-repeated angst among memory enthusiasts that benchmarks never accurately capture real-life performance differences due to the extremely random nature of RAM use. Most benchmarks use a combination of short- and long-length random and streaming access, which is not really how Windows (for example) uses it.
