This paper introduces how to design unified GEMM with unary computing.
Present an ultra-low power unary computing architecture for GEMM, which is compatible for both temporal and rate coding.
General matrix multiplication (GEMM) is universal in various applications, such as signal processing, machine learning, and computer vision. Conventional GEMM hardware architectures based on binary computing exhibit low area and energy efficiency as …
Present the basic concept of unary computing, as well as its mechanism and application.
Present uGEMM architecture, a high performance uanry computing architecture to unify the computation for both rate- and temporal-coded bit streams.
Present division and square root design in stochastic computing, which achieve high accuracy and efficiency by leveraging stochastic correlation among bit streams.
Stochastic Computing (SC) is designed to minimize hardware area and power consumption compared to traditional binary-encoded computation, stemming from the bit-serial data representation and extremely straightforward logic. Though existing Stochastic …
Introduce deep learning and stochastic computing, as well as the latest research overlapping both.
Present design of nonlinear functions for stochastic computing, which can be used in Capsule Neural Network.
An emerging computing scheme based on unary bit streams.