This paper proposes a Matlab framework for the optimized design of mixed-signal accelerator for Deep Neural Networks (DNNs), based on the Flipped (F)-2T2R RRAM compute cell. The manuscript describes an analytical model including the fundamental sources of accelerator non-ideality, developed for a simplified yet accurate system description. The framework allows to explore the design space and optimize the accelerator, targeting system level performance. A trade-off between the accelerator energy consumption and the maximum SNR emerges from simulations.
An Optimization Framework for Mixed-Signal Accelerators Based on F-2T2R Compute Cells / Caselli, Michele; Boni, Andrea. - (2024), pp. 1-4. (Intervento presentato al convegno 2024 20th International Conference on Synthesis, Modeling, Analysis and Simulation Methods and Applications to Circuit Design (SMACD) tenutosi a Volos, Grecia nel 2-5 July 2024) [10.1109/smacd61181.2024.10745462].
An Optimization Framework for Mixed-Signal Accelerators Based on F-2T2R Compute Cells
Caselli, Michele;Boni, Andrea
2024-01-01
Abstract
This paper proposes a Matlab framework for the optimized design of mixed-signal accelerator for Deep Neural Networks (DNNs), based on the Flipped (F)-2T2R RRAM compute cell. The manuscript describes an analytical model including the fundamental sources of accelerator non-ideality, developed for a simplified yet accurate system description. The framework allows to explore the design space and optimize the accelerator, targeting system level performance. A trade-off between the accelerator energy consumption and the maximum SNR emerges from simulations.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
