Performance Analysis of a Phase-Change Memory System on Various CNN Inference Workloads

Jihoon Jang; Hyun Kim; Hyokeun Lee

doi:10.1109/ISOCC56007.2022.10031496

Performance Analysis of a Phase-Change Memory System on Various CNN Inference Workloads

Jihoon Jang
, Hyun Kim
, Hyokeun Lee

Dept. of Electrical and Information Engineering

Seoul National University of Science and Technology (SNUST)
Seoul National University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

6 Scopus citations

Abstract

In this paper, we analyze the suitability of convolutional neural network (CNN) inference workloads on a phase-change memory (PCM) platform. CNN inference has an average of 14\times more read requests than write requests (i.e., read dominant) and a significantly low last-level cache misses per kilo instructions (LLC MPKI) of 2 on average (i.e., computation intensive). In addition, to compare the latency and energy of PCM and DRAM systems, we evaluate CNN inference workloads on two memory systems through a memory simulator. As a result, compared to DRAM, PCM can save total energy by 54% on average, but instruction per cycle (IPC) of PCM is reduced by an average of 28%. In conclusion, CNN inference is a workload suitable for PCM in terms of energy efficiency, but it must be accompanied by a scheme to improve IPC for practical use.

Original language	English
Title of host publication	Proceedings - International SoC Design Conference 2022, ISOCC 2022
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	133-134
Number of pages	2
ISBN (Electronic)	9781665459716
DOIs	https://doi.org/10.1109/ISOCC56007.2022.10031496
State	Published - 2022
Event	19th International System-on-Chip Design Conference, ISOCC 2022 - Gangneung-si, Korea, Republic of Duration: 19 Oct 2022 → 22 Oct 2022

Publication series

Name	Proceedings - International SoC Design Conference 2022, ISOCC 2022

Conference

Conference	19th International System-on-Chip Design Conference, ISOCC 2022
Country/Territory	Korea, Republic of
City	Gangneung-si
Period	19/10/22 → 22/10/22

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 7 Affordable and Clean Energy

Keywords

convolution neural networks
memory simulation
non-volatile memory
Phase-change memory

Access to Document

10.1109/ISOCC56007.2022.10031496

Cite this

Jang, J., Kim, H., & Lee, H. (2022). Performance Analysis of a Phase-Change Memory System on Various CNN Inference Workloads. In Proceedings - International SoC Design Conference 2022, ISOCC 2022 (pp. 133-134). (Proceedings - International SoC Design Conference 2022, ISOCC 2022). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ISOCC56007.2022.10031496

@inproceedings{01de4baa1dee48428fe503aa4cfa032d,

title = "Performance Analysis of a Phase-Change Memory System on Various CNN Inference Workloads",

abstract = "In this paper, we analyze the suitability of convolutional neural network (CNN) inference workloads on a phase-change memory (PCM) platform. CNN inference has an average of 14\textbackslash{}times more read requests than write requests (i.e., read dominant) and a significantly low last-level cache misses per kilo instructions (LLC MPKI) of 2 on average (i.e., computation intensive). In addition, to compare the latency and energy of PCM and DRAM systems, we evaluate CNN inference workloads on two memory systems through a memory simulator. As a result, compared to DRAM, PCM can save total energy by 54\% on average, but instruction per cycle (IPC) of PCM is reduced by an average of 28\%. In conclusion, CNN inference is a workload suitable for PCM in terms of energy efficiency, but it must be accompanied by a scheme to improve IPC for practical use.",

keywords = "convolution neural networks, memory simulation, non-volatile memory, Phase-change memory",

author = "Jihoon Jang and Hyun Kim and Hyokeun Lee",

note = "Publisher Copyright: {\textcopyright} 2022 IEEE.; 19th International System-on-Chip Design Conference, ISOCC 2022 ; Conference date: 19-10-2022 Through 22-10-2022",

year = "2022",

doi = "10.1109/ISOCC56007.2022.10031496",

language = "English",

series = "Proceedings - International SoC Design Conference 2022, ISOCC 2022",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "133--134",

booktitle = "Proceedings - International SoC Design Conference 2022, ISOCC 2022",

}

Jang, J, Kim, H & Lee, H 2022, Performance Analysis of a Phase-Change Memory System on Various CNN Inference Workloads. in Proceedings - International SoC Design Conference 2022, ISOCC 2022. Proceedings - International SoC Design Conference 2022, ISOCC 2022, Institute of Electrical and Electronics Engineers Inc., pp. 133-134, 19th International System-on-Chip Design Conference, ISOCC 2022, Gangneung-si, Korea, Republic of, 19/10/22. https://doi.org/10.1109/ISOCC56007.2022.10031496

Performance Analysis of a Phase-Change Memory System on Various CNN Inference Workloads. / Jang, Jihoon; Kim, Hyun; Lee, Hyokeun.
Proceedings - International SoC Design Conference 2022, ISOCC 2022. Institute of Electrical and Electronics Engineers Inc., 2022. p. 133-134 (Proceedings - International SoC Design Conference 2022, ISOCC 2022).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Performance Analysis of a Phase-Change Memory System on Various CNN Inference Workloads

AU - Jang, Jihoon

AU - Kim, Hyun

AU - Lee, Hyokeun

PY - 2022

Y1 - 2022

N2 - In this paper, we analyze the suitability of convolutional neural network (CNN) inference workloads on a phase-change memory (PCM) platform. CNN inference has an average of 14\times more read requests than write requests (i.e., read dominant) and a significantly low last-level cache misses per kilo instructions (LLC MPKI) of 2 on average (i.e., computation intensive). In addition, to compare the latency and energy of PCM and DRAM systems, we evaluate CNN inference workloads on two memory systems through a memory simulator. As a result, compared to DRAM, PCM can save total energy by 54% on average, but instruction per cycle (IPC) of PCM is reduced by an average of 28%. In conclusion, CNN inference is a workload suitable for PCM in terms of energy efficiency, but it must be accompanied by a scheme to improve IPC for practical use.

AB - In this paper, we analyze the suitability of convolutional neural network (CNN) inference workloads on a phase-change memory (PCM) platform. CNN inference has an average of 14\times more read requests than write requests (i.e., read dominant) and a significantly low last-level cache misses per kilo instructions (LLC MPKI) of 2 on average (i.e., computation intensive). In addition, to compare the latency and energy of PCM and DRAM systems, we evaluate CNN inference workloads on two memory systems through a memory simulator. As a result, compared to DRAM, PCM can save total energy by 54% on average, but instruction per cycle (IPC) of PCM is reduced by an average of 28%. In conclusion, CNN inference is a workload suitable for PCM in terms of energy efficiency, but it must be accompanied by a scheme to improve IPC for practical use.

KW - convolution neural networks

KW - memory simulation

KW - non-volatile memory

KW - Phase-change memory

UR - https://www.scopus.com/pages/publications/85148451120

U2 - 10.1109/ISOCC56007.2022.10031496

DO - 10.1109/ISOCC56007.2022.10031496

M3 - Conference contribution

AN - SCOPUS:85148451120

T3 - Proceedings - International SoC Design Conference 2022, ISOCC 2022

SP - 133

EP - 134

BT - Proceedings - International SoC Design Conference 2022, ISOCC 2022

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 19th International System-on-Chip Design Conference, ISOCC 2022

Y2 - 19 October 2022 through 22 October 2022

ER -

Performance Analysis of a Phase-Change Memory System on Various CNN Inference Workloads

Abstract

Publication series

Conference

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this