
The QX6700 is based on the same 65nm manufacturing process as the other Core 2 processors. It incorporates a total of 8MB of L2 cache, as 4MB per pair of cores. Therein lies a problem: the two caches aren't interconnected. So each pair of cores shares 4MB of cache, and if processor A of core-pair X needs something from the cache of core-pair Y, it has to go out to memory. Juggling cache access to make sure the four cores don't get out of sync with each other is no trivial task.
With the current architecture this will remain an issue, but Intel may solve this when it implements a single-die quad-core chip. But as it stands now, you won't see the full potential of the quad-core processor, as you'll see from some of the benchmark numbers on the following pages. With current and even future operating system support, the advantage of moving from two cores to four doesn't compare to the leap from the NetBurst architecture to Core 2.
This doesn't mean that there isn't a performance increase if you use the right type of multi-threaded applications. If you're into graphics rendering, in particular, you'll see a large performance increase. However, the QX6700's lower clock speed, 2.66GHz, means that in applications that can't take advantage of all the cores, you'll see it fall behind the older Core 2 Extreme X6800, clocked at 2.93GHz.