Memory bandwidth and latency analysis
Interestingly, the lower-latency G.SKILL pack comes out on top in memory bandwidth, according to our ScienceMark 2.0 and Sandra results, albeit my insignificant margins once the tests' run-to-run variance is taken into account.
The problem that faces the Crucial set, and to a slightly lesser extent the G.SKILL, is the lack of northbridge-to-CPU bandwidth, which totals some 12.8GiB/s to the 1.6GHz FSB-clocked QX9770 processor, whereas the modules can deliver, potentially, 25.6GiB/s and 32GiB/s for the G.SKILL and Crucial sets in dual-channel mode.
That's why the extra bandwidth is wasted; you really, really need to have an astronomically-high FSB speed to make use of 1.6GHz+ DDR3.
Latency analysis shows that both sets are on a par, so what the G.SKILL loses in pure frequency is hidden by its 7-7-7-18 timings.
