Sunnyvale (CA) – AMD’s quad-core architecture won’t be released until mid-2007, but the company is already discussing some features of the new Opterons and Athlons. Power consumption will remain a center piece of AMD’s product strategy: Clock speed control of the individual cores will allow the chips to remain in the same power envelope as their dual-core predecessors.
About a year ago, dual-core processors were just about becoming available to the enthusiast computer buyer. Within another year, a dual-core system will be making its way deep into the entry-level segment of the desktop and notebook PC market. By then, quad-core systems are expected to dominate the high-end range of PC and x86 server systems.
Intel has been touting its quad-core processors “Kentsfield,” likely to be named Core 2 Quadro, and “Clovertown,” a future member of the Xeon DP 5000 series, since the beginning of this year and recently announced that it will be releasing Kentsfield as high-end desktop CPU in time for the Christmas season.
AMD will counter Kentsfield and Clovertown with a desktop and a new Opteron chip, but details are scarce. What we learned earlier is that the first quad-core, likely the Opteron variant, will launch in mid-2007 and that it will be a 65 nm chip, with a 45 nm version likely to follow in the first half of 2008. During a recent presentation, AMD provided a few more pieces of information on the architecture of the processor and how it will be able to reach a 68 watt power envelope.
While clock speeds have not been revealed, each of the four cores will integrate 64 kB L1 Cache and 512 kB L2 cache. The native quad-core architecture will also include a 2 MB shared L3 cache, which may increase in capacity over time. The processor will have a total of four Hypertransport links – up from three today – that provide a total bandwidth to outside devices of 5.2 GB/s. AMD is also thinking about integrating support for FB-DIMMs “when appropriate.”
AMD’s startegy to reduce quad-core power consumption: Control every core
But it appears that pure speed isn’t so much the focus of the quad-cores, but that AMD is determined to regain its unquestioned leadership in power-efficiency with its quad-core processors. While Intel’s Kentsfield and Clovertown processors are expected to bring a substantial increase in power consumption to a thermal designer power rating of at least 110 watts, AMD claims that its quad-cores will not consume more than today’s processors.
The key to achieve this goal is AMD’s single-die architecture and its ability to individually adjust the clock speed of each processor core. For example, if the full processing power of all four cores isn’t needed, the architecture is able to reduce the clock speed of individual cores. One core running at full speed and three cores at one third of their maximum clock speed would drop power consumption by 40%. AMD can even completely shut down individual cores for even greater reduction of the CPU’s power consumption.
Compared to AMD’s single-die architecture, Intel’s Kentsfield and Clovertown will be tow-die processors built from two Core 2 Duo chips and two Xeon 5100 processors, which won’t allow Intel to individually adjust clock speeds. Intel has not yet confirmed an increase in power consumption, but mentioned that it sees quad-core processors to remain on the very high-end of desktop computing for some time. And since gamers usually could care less how much power their computer system uses, AMD may just have enough headroom left to catch up with Intel’s performance once the quad-core CPUs are released.