This new architecture will act as the base for Intel’s Xeon and Core series moving forward and includes 3D architecture. Called ‘Foveros’, this allows Intel to stack graphics and AI processors on top of its CPUs. Previously 2D architecture featured components that were next to each other communicating via silicon bridges, but with 3D architecture, the components can be stacked on top of each other which comes with a number of performance and power saving benefits.
We’re also going to see algorithms introduced to greatly improve memory latency too, and the whole package will allow more processes to be completed at the same time.
Why is smaller processor architecture better?
Jumps in nano-architecture development come with a number of benefits – a smaller product means less distance for the electricity to travel across the processor’s paths. This not only allows information to travel faster, but it also requires less power to move that information and so generates less heat. Heat is one of the main factors in limiting processor clock speed, as while you can overclock a processor by just driving more power into it – it’s not particularly useful if it explodes afterwards.
Smaller processors can be manufactured with less silicon too, which means that if a piece of silicon has some faults within it, each processor taking up a smaller percentage of the silicon will result in a smaller percentage of processors with faults - this drives down costs in multiple ways, which means making processors becomes cheaper and better for everyone (but mostly Intel).