Feasibility study of the Software-Based Self Test methodology on a Hardware Accelerator for Neural Networks.-

Lorenzo Zaia

Feasibility study of the Software-Based Self Test methodology on a Hardware Accelerator for Neural Networks.-.

Rel. Paolo Bernardi. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Informatica (Computer Engineering), 2020

Preview

PDF (Tesi_di_laurea) - Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives.
Download (2MB) | Preview

Abstract

The general subject of the MS Thesis is the preliminary study about the reliability issues of Hardware Accelerator for Neural Networks and the counteractions that can be taken to detect erroneous behaviours occurring during the mission of the device.