Mixed-precision uses in each layer of a Deep Neural Network the minimum bit-width that preserves accuracy. In this context, our new Reconfigurable 2D-Convolution Module (RCM) computes N =1, 2 or 4 Multiply-and-Accumulate operations in parallel with configurable precision from 1 to 16/N bits. Our design-space exploration via high-level synthesis obtains the best points in the latency vs area space, varying the size of the tensor tile handled by our RCM and its parallelism. A comparison with a non-configurable module on a 28-nm technology shows many reconfigurable Pareto points for low bit-width configurations, making our RCM a promising mixed-precision accelerator for inference.

A Reconfigurable 2D-Convolution Accelerator for DNNs Quantized with Mixed-Precision / Casu, Mario Roberto; Urbinati, Luca. - ELETTRONICO. - (2023), pp. 210-215. (Intervento presentato al convegno Applepies 2022 tenutosi a Genova nel September 26–27, 2022) [10.1007/978-3-031-30333-3_27].

A Reconfigurable 2D-Convolution Accelerator for DNNs Quantized with Mixed-Precision

Casu, Mario Roberto;Urbinati, Luca
2023

Abstract

Mixed-precision uses in each layer of a Deep Neural Network the minimum bit-width that preserves accuracy. In this context, our new Reconfigurable 2D-Convolution Module (RCM) computes N =1, 2 or 4 Multiply-and-Accumulate operations in parallel with configurable precision from 1 to 16/N bits. Our design-space exploration via high-level synthesis obtains the best points in the latency vs area space, varying the size of the tensor tile handled by our RCM and its parallelism. A comparison with a non-configurable module on a 28-nm technology shows many reconfigurable Pareto points for low bit-width configurations, making our RCM a promising mixed-precision accelerator for inference.
2023
978-3-031-30333-3
978-3-031-30332-6
File in questo prodotto:
File Dimensione Formato  
published_paper.pdf

non disponibili

Tipologia: 2a Post-print versione editoriale / Version of Record
Licenza: Non Pubblico - Accesso privato/ristretto
Dimensione 1.93 MB
Formato Adobe PDF
1.93 MB Adobe PDF   Visualizza/Apri   Richiedi una copia
accepted_version.pdf

Open Access dal 30/04/2024

Tipologia: 2. Post-print / Author's Accepted Manuscript
Licenza: PUBBLICO - Tutti i diritti riservati
Dimensione 5.58 MB
Formato Adobe PDF
5.58 MB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2978332