We love it when we're right. Okay, so it isn't exactly a surprise, but AMD's announcement that its native quad-core Opterons will ship in August means that, as long as the company hits that ship date, we don't have to update the old CPU road map feature. What it means for you is less certain.
If AMD's plans stay on track, the quad-core tech that comes late this summer to AMD's server/workstation Opterons should roll into AMD's forthcoming Phenom desktop and laptop CPUs later this year. And on paper, AMD could have a speed advantage over Intel's Core 2 Quad Q and Core 2 Extreme QX quad-core chips, due to the fact that AMD's design puts four cores on a single piece of silicon, rather than Intel's current two-and-two design. However, the news that the first quad-core Opterons will ship with a max speed of 2.0GHz makes us wonder just what kind of thermal situation AMD is dealing with. Intel's fastest quad-core Xeon server chip currently runs at 2.66Ghz.
AMD's quad-core design is native, which means it will have a unified L2 cache to pull from, which, we're told, can scale dynamically based on the workload. That means more data bandwidth when the CPU performs one task, and more efficient throughput management with multiple operations going on at once. In short, it could be faster than Intel's design, which gives each pair of cores its own, smaller L2 cache allocation. But whether an advantage in data throughput is enough to counter the benefits of faster core clock speed, we can't say until we see the benchmarks. We expect that on both the server and desktop sides AMD will bump up the clocks as much as it can, but, of course, Intel isn't standing pat, and its move to a more heat-efficient 45nm manufacturing process may let Intel ramp up clock speeds even beyond AMD's forthcoming 65nm parts.