Characterizing Memory Access Patterns of Various Convolutional Neural Networks for Utilizing Processing-in-Memory

Jihoon Jang, Hyun Kim, Hyokeun Lee

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

4 Scopus citations

Abstract

Convolutional neural network (CNN) models require deeper networks and more training data for better performance, which in turn results in greater computational and memory requirements. In this paper, we analyze the memory access patterns that occur in main memory during the training processes of various CNN models. CNN training is a linear procedure consisting of a forward pass (FP) and a backward pass (BP). As a result of the analysis, we found that BP accounted for 83.4% of the total main memory accesses on average. Therefore, CNN training including FP and BP is much more memory-intensive than CNN inference using only FP. This demonstrates that CNN training is a suitable application for near-data processing to reduce memory bottlenecks and conserve computational resources.

Original languageEnglish
Title of host publication2023 International Conference on Electronics, Information, and Communication, ICEIC 2023
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9798350320213
DOIs
StatePublished - 2023
Event2023 International Conference on Electronics, Information, and Communication, ICEIC 2023 - Singapore, Singapore
Duration: 5 Feb 20238 Feb 2023

Publication series

Name2023 International Conference on Electronics, Information, and Communication, ICEIC 2023

Conference

Conference2023 International Conference on Electronics, Information, and Communication, ICEIC 2023
Country/TerritorySingapore
CitySingapore
Period5/02/238/02/23

Keywords

  • Convolutional Neural Networks
  • Memory access pattern
  • Near data processing
  • Network training
  • Processing-in-Memory

Fingerprint

Dive into the research topics of 'Characterizing Memory Access Patterns of Various Convolutional Neural Networks for Utilizing Processing-in-Memory'. Together they form a unique fingerprint.

Cite this