Uncover AMD's future oriented small chip design

The core idea of small chip, also translated as core particle, is to reduce the time and cost of chip development by directly integrating the pre developed and designed bare chip into IC packaging.

At present, the mainstream approach is that if you want to build high-performance chips, you need to develop system level chips (SOC), and then cooperate with the advanced technology of the OEM to miniaturize the internal functional units and use chips in the same area or smaller area to achieve higher performance. Obviously, as the feature size gradually approaches the physical limit and the process complexity continues to increase, this path becomes more and more difficult.

Generally speaking, the practice of small chip is to build the chip like a building block. AMD, Intel, TSMC, Marvell, cadence and other industry giants are quite concerned about this and regard it as one of the options to continue Moore's law. Here, let's take a look at AMD's layout in terms of small chips.

It can be said that today, when the industry pays attention to small chips, amd has played an important role. The concept of small chip can be traced back to the multi chip module in the 1970s - the original chip is made and then assembled. In 2014, chip design companies began to pay attention to this technology. In 2016, DARPA, the Defense Advanced Research Projects Agency, launched the chips project, which mentioned the idea of chip reuse. But the small chip is really famous because of the success of AMD EYPC Series CPU.

In 2017, amd developed the EYPC server processor "Naples" using a small chip in its "Zen 2" architecture. According to AMD engineers at that time, the processor adopted such an innovative method to reduce costs. According to AMD engineers at that time, compared with system on chip design, this innovative method reduced the cost by half and greatly shortened the design time. Subsequently, amd deployed "Zen 2" small chip technology in consumer CPU and enterprise epyc processor.