Page 1 of 5Introduction
When DDR2 was first announced, everyone was skeptical about its performance. It was capable of scaling to extremely high frequencies, but it also brought along high latencies that enthusiasts dread. So what's good about using DDR2, other than the higher frequencies? The new TCCD memory is known to be able to scale above 270MHZ with reasonable timings. Do we really need to go beyond 300MHZ and be lagged by high latencies? Fortunately and theoretically, DDR2 is supposed to carry twice as much data per cycle than regular DDR. Hence in lamen's terms, DDR2 that carries twice the load with latencies of 4-4-4-12 should equal to the performance of DDR that carries a single load at 2-2-2-6 latencies. Unfortunately this is not the case.
The initial release of DDR2 was marked with high latencies, usually in the CAS 5 area. Corsair was the first to announce the "ultra low" latencies of high frequency DDR2 with their release of the XMS2-5400C4PRO. The timings were rated at 4-4-4-12, which is not all that low. However for DDR2, it's considered to be pretty good. Kingston quickly joined the games and produced the new HyperX2 line of memory. These modules had even lower latencies, running at 4-4-4-10. Although it's not much lower, it is still lower.