Increasing Throughput of In-Memory DNN Accelerators by Flexible Layerwise DNN Approximation.

Research output: Contribution to journalResearch articleContributedpeer-review

Contributors

  • Cecilia De la Parra - , Robert Bosch GmbH (Author)
  • Taha Soliman - , Robert Bosch GmbH (Author)
  • Andre Guntoro - , Robert Bosch GmbH (Author)
  • Akash Kumar - , Chair of Processor Design (cfaed) (Author)
  • Norbert Wehn - , University of Kaiserslautern-Landau (Author)

Abstract

Approximate computing and mixed-signal in-memory accelerators are promising paradigms to significantly reduce computational requirements of deep neural network (DNN) inference without accuracy loss. In this work, we present a novel in-memory design for layerwise approximate computation at different approximation levels. A sensitivity-based high-dimensional search is performed to explore the optimal approximation level for each DNN layer. Our new methodology offers high flexibility and optimal tradeoff between accuracy and throughput, which we demonstrate by an extensive evaluation on various DNN benchmarks for medium- and large-scale image classification with CIFAR10, CIFAR100, and ImageNet. With our novel approach, we reach an average of 5× - and up to 8× - speedup without accuracy loss.

Details

Original languageEnglish
Pages (from-to)17-24
Number of pages8
JournalIEEE Micro
Volume42
Issue number6
Publication statusPublished - 2022
Peer-reviewedYes

External IDs

Scopus 85135767385

Keywords

Research priority areas of TU Dresden

Library keywords