Special Session - Non-Volatile Memories: Challenges and Opportunities for Embedded System Architectures with Focus on Machine Learning Applications

Henkel J, Sidduh L, Bauer L, Teich J, Wildermann S, Tahoori MB, Mayahinia M, Castrillon J, Khan AA, Farzaneh H, de Lima JPC, Chen JJ, Hakert C, Chen KH, Yang CL, Cheng HY (2023)

Publication Language: English

Publication Type: Conference contribution, Original article

Publication year: 2023

Conference Proceedings Title: Proceedings of the International Conference on Compilers, Architectures, and Synthesis for Embedded Systems (CASES)

Event location: HAMBURG DE

Abstract

This paper explores the challenges and opportunities of integrating
non-volatile memories (NVMs) into embedded systems for ma-
chine learning. NVMs offer advantages such as increased memory
density, lower power consumption, non-volatility, and compute-in-
memory capabilities. The paper focuses on integrating NVMs into
embedded systems, particularly in intermittent computing, where
systems operate during periods of available energy. NVM technolo-
gies bring persistence closer to the CPU core, enabling efficient
designs for energy-constrained scenarios. Next, computation in re-
sistive NVMs is explored, highlighting its potential for accelerating
machine learning algorithms. However, challenges related to relia-
bility and device non-idealities need to be addressed. The paper also
discusses memory-centric machine learning, leveraging NVMs to
overcome the memory wall challenge. By optimizing memory lay-
outs and utilizing probabilistic decision tree execution and neural
network sparsity, NVM-based systems can improve cache behavior
and reduce unnecessary computations. In conclusion, the paper
emphasizes the need for further research and optimization for the
widespread adoption of NVMs in embedded systems presenting
relevant challenges, especially for machine learning applications.

Authors with CRIS profile

Jürgen Teich Lehrstuhl für Informatik 12 (Hardware-Software-Co-Design) Stefan Wildermann Lehrstuhl für Informatik 12 (Hardware-Software-Co-Design)

Related research project(s)

Automatic Cross-Layer Synthesis of High Performance, (Ultra-)Low Power Hardware Implementations from Data Flow Specifications by Integration of Emerging FeFET Technology (HiLoDa Nets) March 1, 2024 - March 1, 2027 HYPNOS – Co-Design of Persistent, Energy-efficient and High-speed Embedded Processor Systems with Hybrid Volatility Memory Organisation (HYPNOS) DFG-Schwerpunktprogramm (SPP) 2377 - Disruptive Speichertechnologien Sept. 21, 2022 - Sept. 21, 2025

Involved external institutions

University of Twente NL Netherlands (NL) Karlsruhe Institute of Technology (KIT) DE Germany (DE) National Taiwan University (NTU) / 國立臺灣大學 TW Taiwan (TW) Technische Universität Dortmund DE Germany (DE) Technische Universität Dresden DE Germany (DE) Academia Sinica / 中央研究院 TW Taiwan (TW)

How to cite

APA:

Henkel, J., Sidduh, L., Bauer, L., Teich, J., Wildermann, S., Tahoori, M.B.,... Cheng, H.-Y. (2023). Special Session - Non-Volatile Memories: Challenges and Opportunities for Embedded System Architectures with Focus on Machine Learning Applications. In Proceedings of the International Conference on Compilers, Architectures, and Synthesis for Embedded Systems (CASES). HAMBURG, DE.

MLA:

Henkel, Jörg, et al. "Special Session - Non-Volatile Memories: Challenges and Opportunities for Embedded System Architectures with Focus on Machine Learning Applications." Proceedings of the International Conference on Compilers, Architectures, and Synthesis for Embedded Systems (CASES), HAMBURG 2023.

BibTeX: Download