Matrix-Vector Multiplication (MVM) is a cornerstone operation in various computational fields, from artificial intelligence to cryptography. In recent years, In-Memory Computing (IMC) has become dominant in the field of energy-efficient computing for data-intensive operations like MVM. In this paper, we propose RRAM IMC capable of performing low-precision multi-bit MVM in a single cycle using novel data mapping strategy and analog carry propagation technique. Our proposed architecture and mapping strategy achieves 21.45 TOPS/W at 65nm technology node simulations.
No takes yet. Share an insight, caveat, or question.
J et al. (2024) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: