Deep Mapper: A Multi-Channel Single-Cycle Near-Sensor DNN Accelerator

Mehrdad Morsali, Sepehr Tabrizchi, Maximilian Liehr, Nathaniel Cady, Mohsen Imani, Arman Roohi, Shaahin Angizi

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This work proposes the Deep Mapper, as a new near-sensor resistive accelerator architecture for Deep Neural Networks (DNN) inference that co-integrates the sensing and computing phases of resource-constrained edge devices. Deep Mapper is developed to intrinsically realize highly parallelized multi-channel processing of input frames supported by a new dense hardware-friendly mapping methodology. Our circuit-to-application simulation results on the DNN acceleration task show that Deep Mapper reaches an efficiency of 4.71 TOp/s/W outperforming state-of-the-art near-/in-sensor accelerators.

Original languageEnglish (US)
Title of host publication2023 IEEE International Conference on Rebooting Computing, ICRC 2023
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9798350382044
DOIs
StatePublished - 2023
Event8th IEEE International Conference on Rebooting Computing, ICRC 2023 - San Diego, United States
Duration: Dec 5 2023Dec 6 2023

Publication series

Name2023 IEEE International Conference on Rebooting Computing, ICRC 2023

Conference

Conference8th IEEE International Conference on Rebooting Computing, ICRC 2023
Country/TerritoryUnited States
CitySan Diego
Period12/5/2312/6/23

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence
  • Hardware and Architecture

Fingerprint

Dive into the research topics of 'Deep Mapper: A Multi-Channel Single-Cycle Near-Sensor DNN Accelerator'. Together they form a unique fingerprint.

Cite this