VNWFET-Based Systolic Array Accelerator for Deep Neural Networks

Vertical Nanowire Field Effect Transistor (VNWFET) is an emerging technology that allows 3D integration thus reducing footprint while maintaining energy-efficient design. In this work, we present a complete toolchain from device to complex logic block synthesis. We developed a VNWFET standard cell library which is based on a real device compact model. This allowed us to successfully synthesize our Neural Network Compute Cube (N
²
C
²
); the basic computational unit of the targeted Systolic Array as well as evaluating the design, performance and cost of the N
²
C
²
based on vertical technology.