Overclocking and performance
The Dominators run at relatively high clock speeds by default, but we wondered if it might be possible to push them even harder. Using our EVGA nForce 680i SLI motherboard, we were able to crank the TWIN2X2048-8888C4DF modules up from a stock speed of 1111MHz to a whopping 1245MHz with 4-4-4-12 timings. The DIMMs were perfectly stable at that speed with a 2T command rate, even under load, although the auxiliary cooling fan was required. We tried pushing the 8888C4DFs beyond 1245MHz, but even with extra voltage, they weren't stable under load and would often refuse to boot Windows.

Our TWIN2X2048-9136C5D modules weren't quite as happy at higher speeds, only hitting 1200MHz before instability reared its ugly head. The DIMMs were solid running at 1200MHz with their default voltage, though, and they didn't need the Dominator cooler.

Overclocking success is never guaranteed, so your mileage may vary. Still, we couldn't resist throwing a few benchmarks at our DIMM configurations to gauge performance. We included a set of Corsair TWIN2X2048-8500C5 modules running at 800MHz and 4-4-4-12 timings for reference, since that's what we've been using for all our Core 2 motherboard reviews. We also have results for the 9136C5Ds at their default 1143MHz and overclocked to 1200MHz. Scores are in for the 8888C4Ds at 1067, 1143, 1200, and 1245MHz, all with 4-4-4-12 timings.

Surprisingly, the 8888C4D's tighter timings don't give it much of an edge in Sandra or Cachemem's bandwidth tests. Turning up the clock speed definitely improves performance, though.

Things get a little more interesting in Cachemem's latency test, where tighter timings definitely lower access times. Higher clock speeds help, too. Nanoseconds might not seem like much, but with today's processors chewing through billions of instructions per second, a nanosecond is a virtual eternity inside a modern PC.

Of course, quicker memory latencies and greater bandwidth only improve application performance if that application is memory-bound. Many are not, and as our Quake 4 timedemo neatly illustrates, bottlenecks often lie elsewhere.

Noise levels
Noise levels were measured with an Extech 407727 Digital Sound Level meter four inches from the top of our open test bench with and without the Dominator's auxiliary cooler.

Even with the Dominator cooler spinning at full speed—unimpeded by automatic or manual fan speed throttling—noise levels are only a decibel and a half louder than without the fan installed. Whether you'll notice the difference very much depends on just how quiet your system is to start, but exceptionally low noise levels and extreme overclocking rarely go hand in hand.