Mixed Circuit of Binary & Stochastic Computing

Duration: Jan. 2022 – Sep. 2023 Advisor: Prof. Weikang Qian at SJTU

Overview

Developed a comprehensive framework for designing and implementing mixed binary-stochastic computing circuits, bridging conventional and stochastic computing paradigms.

Key Contributions

• Defined a domain-specific language (DSL) for agile mixed circuit development
• Implemented a Verilog code generation backend for the DSL to facilitate hardware synthesis
• Evaluated mixed binary-stochastic circuits for image processing applications

Publications

This work resulted in the SCGen framework, published at DATE 2024, providing a versatile generator framework for stochastic circuit development.

Impact

Enabled rapid prototyping and evaluation of novel computing paradigms that combine the benefits of both binary and stochastic computing approaches.