Inside the processor there has been an increased concentration of cores (number of CPU's on a single chip). So the effect of more cores on a single chip mimics the physical presence of two or more processors requiring lesser energy because of the unique architecture and providing more efficiency. The latest processors have four cores, but manufacturers like Intel and AMD have released roadmaps for octal core processors stated for release in 2009 thereby setting the stage for 12, 16, and 20... processors on a single integrated circuit.

With all these cores in a tiny silver of silicon, networking and data transfer between the cores in the die (single integrated circuit) will need to have a quantum change. Unlike today's processor architecture where running different programs and applications means that the processor switches rapidly between them, future designs with say a hundred cores will have separate cores dedicated to individual processes. This means that word processing would be taken care of by one core, graphics processing by another, running a game on the third and so on. In such a scenario, current bus transfers data rates (transfer of data from one part of the chip to the other) simply cannot keep up. Going by the current architecture the data transfer rate would be lagging behind so badly that it would mean that most of the cores would be starved of data, and instead of speeding up individual applications they would slow down considerably.

To get around the inter-chip networking bottlenecks, companies like Intel and Rambus have been working on prototypes of newer architectural models of chip design. Rambus has a model called Terabyte Bandwidth Initiative (TBI) and Intel's programme is called Terascale Computing Research Program. Intel has released an eighty core chip that delivers super computer grade speeds on desktop computers. The finger-nail sized processor's performance is 1.81 trillion floating point calculations per second at the speed of 5.7 gigahertz which is about three hundred times faster than the fastest processors available in the market today. Data transfer rates achieved by these models have peak rates of up to 2.92 terabits per second which is approximately 374 gigabytes per second, putting them in numbers fifteen hundred full length movies per second.

How is all this going to be possible? In Intel's case each core has a 5-port message passing mesh network which is connected in a two dimensional mesh network with other cores that implement message passing- an efficient local area network on a chip. The cores will communicate with each better and a router will ensure that the right data goes to the right chip. Think of it like every chip knows what the other is doing and shares the workload because of the excellent communications network. This architecture is a lot more scalable than present day multi-core technologies in terms of speed and interconnectivity. The Rambus network differs in the sense that instead of sending one message per wire it has the capacity to send multiple messages per wire resulting in faster transfer. These technologies are stated to come in the market by 2010 according to the companies' statements.