Approximate computing is a well-established technique to mitigate power consumption in error-tolerant domains such as image processing and machine learning. When paired with reconfigurable hardware, it enables dynamic adaptability to each specific task with improved power-accuracy trade-offs. In this work, we present a design methodology to enhance the energy and error metrics of a signed multiplier. This novel approach reduces the approximation error by leveraging a statistic-based truncation strategy. Our multiplier features 256 dynamically configurable approximation levels and run-time selection of the result precision. Our technique improves the mean-relative error by up to 34% compared to the zero truncation mechanism. Compared with an exact design, we achieve a maximum of 60.1% power saving for a PSNR of 10.3dB on a 5x5 Sobel filter. Moreover, we reduce the computation energy of LeNet by 31.5%, retaining 89.4% of the original accuracy on FashionMNIST.

TEMET: Truncated REconfigurable Multiplier with Error Tuning / Guella, Flavia; Valpreda, Emanuele; Caon, Michele; Masera, Guido; Martina, Maurizio. - ELETTRONICO. - 1110:(2024), pp. 370-377. (Intervento presentato al convegno International Conference on Applications in Electronics Pervading Industry, Environment and Society tenutosi a Genova, Italy nel 28-29 September 2023) [10.1007/978-3-031-48121-5_53].

TEMET: Truncated REconfigurable Multiplier with Error Tuning

Guella, Flavia;Valpreda, Emanuele;Caon, Michele;Masera, Guido;Martina, Maurizio
2024

Abstract

Approximate computing is a well-established technique to mitigate power consumption in error-tolerant domains such as image processing and machine learning. When paired with reconfigurable hardware, it enables dynamic adaptability to each specific task with improved power-accuracy trade-offs. In this work, we present a design methodology to enhance the energy and error metrics of a signed multiplier. This novel approach reduces the approximation error by leveraging a statistic-based truncation strategy. Our multiplier features 256 dynamically configurable approximation levels and run-time selection of the result precision. Our technique improves the mean-relative error by up to 34% compared to the zero truncation mechanism. Compared with an exact design, we achieve a maximum of 60.1% power saving for a PSNR of 10.3dB on a 5x5 Sobel filter. Moreover, we reduce the computation energy of LeNet by 31.5%, retaining 89.4% of the original accuracy on FashionMNIST.
2024
978-3-031-48121-5
File in questo prodotto:
File Dimensione Formato  
temet_final.pdf

embargo fino al 03/03/2025

Descrizione: Versione autore
Tipologia: 2. Post-print / Author's Accepted Manuscript
Licenza: PUBBLICO - Tutti i diritti riservati
Dimensione 337.91 kB
Formato Adobe PDF
337.91 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
Applepies_2023.pdf

non disponibili

Descrizione: Versione editorial
Tipologia: 2a Post-print versione editoriale / Version of Record
Licenza: Non Pubblico - Accesso privato/ristretto
Dimensione 337.91 kB
Formato Adobe PDF
337.91 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
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/11583/2983644