Master's Thesis

A RISC-V Execution and Control System for Photonic Acceleration of AI Applications

Ricardo José Luzio Moura — 2026

Key information

Authors:

Ricardo José Luzio Moura (Ricardo José Luzio Moura)

Supervisors:

Rui Silva Ferreira (Rui Silva Ferreira); Luís Jorge Brás Monteiro Guerra e Silva (Luís Jorge Brás Monteiro Guerra e Silva)

Published in

June 30, 2026

Abstract

The increasing computational demands of modern Artificial Intelligence workloads have driven the development of hardware accelerators. Photonic accelerators have emerged as a promising solution, offering significant advantages in energy efficiency and computational speed over conventional digital implementations. However, to achieve this potential, it is necessary to design and implement a software infrastructure to interface with the Photonic Accelerator. This thesis presents the design and implementation of a software environment that enables the control and operation of a photonic hardware accelerator within a RISC-V based System on Chip, developed in the context of the NEUROPULS project. The environment consists of a functional simulator of the accelerator, an Application Programming Interface that abstracts the low-level hardware details, and the hardware execution environment running on the target FPGA platform. This allows use-case applications developed by project partners to be tested independently of the hardware availability, while sharing the same interface across all execution environments. Additionally, two sorting algorithms were developed to minimize weight reconfiguration overhead during batch inference. The Zephyr Real-Time Operating System was also integrated on a Microblaze-V soft processor deployed on the AMD Zynq UltraScale+ RFSoC ZCU208 FPGA platform, including the necessary block designs and configuration steps for a dual-processor architecture combining an ARM Processing System and a RISC-V Programmable Logic core.

Publication details

Authors in the community:

Supervisors of this institution:

Fields of Science and Technology (FOS)

electrical-engineering-electronic-engineering-information-engineering - Electrical engineering, electronic engineering, information engineering

Publication language (ISO code)

eng - English

Rights type:

Embargoed access

Date available:

May 1, 2027

Institution name

Instituto Superior Técnico