A 95.6-TOPS/W Deep Learning Inference Accelerator With Per-Vector Scaled 4-bit Quantization in 5 nm
Sign Up to like & getrecommendations! Published in 2023 at "IEEE Journal of Solid-State Circuits"
DOI: 10.1109/jssc.2023.3234893
Abstract: The energy efficiency of deep neural network (DNN) inference can be improved with custom accelerators. DNN inference accelerators often employ specialized hardware techniques to improve energy efficiency, but many of these techniques result in catastrophic… read more here.
Keywords: per vector; quantization; accuracy loss; accelerator ... See more keywords
A 0.02% Accuracy Loss Voltage-Mode Parallel Sensing Scheme for RRAM-Based XNOR-Net Application
Sign Up to like & getrecommendations! Published in 2022 at "IEEE Transactions on Circuits and Systems II: Express Briefs"
DOI: 10.1109/tcsii.2022.3157767
Abstract: Compute-in-memory (CiM) is an effective way to solve the memory wall issue. As a promising candidate for CiM, resistive random access memory (RRAM) has the advantages of non-volatile, fast operating speed, and programmable resistance. However,… read more here.
Keywords: accuracy loss; rram; inline formula; tex math ... See more keywords
Exposing Reliability Degradation and Mitigation in Approximate DNNs Under Permanent Faults
Sign Up to like & getrecommendations! Published in 2023 at "IEEE Transactions on Very Large Scale Integration (VLSI) Systems"
DOI: 10.1109/tvlsi.2023.3238907
Abstract: Approximate computing is known for enhancing deep neural network accelerators’ energy efficiency by introducing inexactness with a tolerable accuracy loss. However, small accuracy variations may increase the sensitivity of these accelerators toward undesired subtle disturbances,… read more here.
Keywords: permanent faults; mitigation approximate; accuracy loss; fault ... See more keywords