Finding a better way to measure progress in semiconductors

EECS Professors Chenming Hu, Tsu-Jae King Liu, Jeffrey Bokor and Sayeef Salahuddin are featured in an IEEE Spectrum article about efforts to better track and showcase the exponential pace of progress in semiconductor technology – the foundation of computing and communication devices, networks and systems. For many decades, Moore’s Law has been used to gauge this trend with the number of transistors on the most advanced microprocessor chip doubling every two years, thanks to advances that allow for further miniaturization of the transistor. But what happens as physical limits such as the finite size of atoms and the speed of light are approached? Does progress in semiconductor technology cease? As co-inventors of the “FinFET” that enabled the industry to shrink transistors to below 10 nanometers in physical dimension, Hu, Liu and Bokor have the gravitas to advocate for a better industry metric to show that progress in semiconductor technology is limited only by human creativity and ingenuity – as it always has been. In June 2019 they met together with Salahuddin, a pioneer in the development of ferroelectric devices, and colleagues from Stanford University, and came up with a metric they dubbed “LMC” (logic, memory, connection). This new metric takes a more holistic view of technology advancement to enable computing performance to improve at an exponential pace through increases in the densities of logic (computing) devices, memory (information storage) cells, and the density of connections (wiring) between logic and memory devices on a chip. Liu sees the LMC metric as the driver of a new era of innovation in semiconductors.