AMD has shed a little more light on its previously disclosed plans for dual-core processors, announcing that it has completed its design for both server and desktop processors and will ship products in 2005.
Both dual-core processors will be based on the company's AMD64 technology, said Fred Weber, vice president of engineering for AMD's Computation Products Group. That architecture includes 64-bit extensions to the 32-bit x86 instruction set as well as an integrated memory controller that helps AMD move to dual-core designs, he said.
A dual-core chip is simply two separate processors on a single chip. In order to deliver the sharp increases in performance that have characterised the chip industry for over a decade, chip designers have been looking for another way to increase performance without increasing the chip's clock speed and therefore the amount of power it consumes.
The industry appears to have settled on dual-core designs as the way it will keep those performance gains going through the middle part of this decade. Sun and IBM have already released dual-core chips for servers as a way to increase performance without resorting to faster clock speeds and therefore increased power consumption.
Intel recently laid out its dual-core strategy, with plans to release dual-core processors for servers, desktops and notebooks next year. In May, it cancelled plans for the single-core Tejas and Jayhawk processors in favour of dual-core desktop and server chips for which codenames have not yet been released.
The power leakage issues brought on by the switch to the 90nm process generation has accelerated the shift to dual-core designs. But those smaller leaky transistors have also made dual-core chips a more economically feasible proposition because manufacturers can now fit two smaller cores on a single die without decreasing the number of chips cut from a silicon wafer, said Dean McCarron, principal analyst with Mercury Research.
The integrated memory controller and Hypertransport interconnect technology present in the AMD64 architecture allow AMD to produce a similar amount of performance from one dual-core chip as provided by a dual-processor server, Weber said.
AMD's technology is designed to connect up to four Opteron processors without a chipset as required by other processors, McCarron said. This makes it easier to connect the processors on a single piece of silicon, since the connection wires are already built into the chip, he said. "They already had a multi-processing design, they're just changing the physical location of where the processors are," McCarron said.
AMD's dual-core server processors will share a single memory controller, Weber said. This won't create a bottleneck because a server with two Opteron chips, and therefore two memory controllers, already has more than enough memory bandwidth required to run that system, he said. "It's always a juggling act to add a little more processing and a little more memory. Right now, we have plenty of memory and I/O bandwidth, so we're adding processing," Weber said. The dual-core chips will work with current socket technology in motherboards that are rated for the specifications of the dual-core chips, Weber said. A BIOS change will be required, but otherwise the chips will work in the same sockets as single-core Opterons, he said.
AMD will also continue to offer three power ratings for its dual-core server processors, Weber said. The company currently offers chips rated at 30 watts, 55 watts, and 89 watts, and will continue to do so for dual-core chips, he said.
The dual-core chips will be manufactured on AMD's 90nm process technology, Weber said. AMD is currently ramping up its 90nm process technology with the first such chips scheduled to be available later this year. The dual-core server chips will launch in the middle of 2005, while dual-core desktop processors will follow in the second half of 2005, Weber said.