Yanghejian Zhang
A test case for the evaluation of AI-oriented hardware accelerators reliability.
Rel. Matteo Sonza Reorda, Juan David Guerrero Balaguera, Josie Esteban Rodriguez Condia. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Elettronica (Electronic Engineering), 2024
|
Preview |
PDF (Tesi_di_laurea)
- Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives. Download (1MB) | Preview |
Abstract
This article focuses on AI-Oriented Hardware Accelerators Reliability Using NVDLA Deep Learning Accelerator from NVIDIA As an example, the NVDLA Deep Learning Accelerator was designed using zoix and terata max software. There are six main parts in the essay, namely introduction, background, implementation, result, discussion and conclution and appendix. introduction introduces the current problem to be solved, a short summary and the structure of the essay. The background section introduces the composition of Convolutional Neural Network, the hardware composition of NVDLA, the software used in the thesis, and the methodology used in the thesis. The implementation section describes how to generate the corresponding test vectors for integer and floating-point multipliers and the corresponding addition trees.
The results section shows the simulation results for the final test vector
Relatori
Anno Accademico
Tipo di pubblicazione
Numero di pagine
Corso di laurea
Classe di laurea
Aziende collaboratrici
URI
![]() |
Modifica (riservato agli operatori) |
