Sponsored by IEEE

For leadership in development and deployment of chiplet architecture designs for high-performance and adaptive computing

AMD engineered a new design approach intended to keep computers improving at the pace of Moore’s law by disintegrating SoC designs into multiple smaller “chiplets.” This innovation allowed the industry to shift production from monolithic pieces of silicon to an assembly of smaller pieces. One outcome has been faster development of new products, as it is quicker to mix and match modular pieces linked by short data connections than to graft and redesign them into a single new chip. Chiplets are also more cost-effective. In processors made up of chiplets, cutting-edge technology can be reserved for the pieces of a design where the investment will pay off the most. Other chiplets, such as those for I/O, can be made using more established and cheaper process nodes. Smaller pieces of silicon are also inherently less prone to manufacturing defects, resulting in higher yields. The chiplet architecture has nearly halved the manufacturing cost of producing modern processors. Chiplets have brought additional advantages such as greater data bandwidth to memory and higher compute capability than possible with more conventional monolithic designs. What’s more, they can improve efficiency by combining components to allow them to work together more closely. Chiplets have proven to be a disruptive force in the industry and are rapidly becoming the new design and manufacturing norm. Simply put, chiplets make it faster and less costly to flexibly assemble I/O, memory, and compute cores. Chiplets have not only helped propel AMD as a leader in this area, but they’ve also brought much-needed innovation and competition—and even cooperation—across the industry.
AMD is the high-performance and adaptive computing leader, powering the products and services that help solve the world’s most important challenges, headquartered in Santa Clara, California, USA.