NavCim: Comprehensive Design Space Exploration for Analog Computing-in-Memory Architectures
Open Access
- 13 October 2024
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 168-182
Abstract
Analog computing-in-memory (ACiM) technology has shown strong potential for neural network accelerators, addressing von-Neumann performance bottlenecks with in-memory data processing and computation. Understanding the ACiM design space, including its trade-offs and constraints, and systematically and effectively exploring it for optimal performance is essential to turn the promise into a viable product. Recent research demonstrated that multi-objective searches for ACiM architectures with heterogeneous tiles can simultaneously optimize power, performance, and area (PPA), outperforming existing tiled ACiM proposals. In this paper, we propose NavCim, a comprehensive ACiM design space exploration mechanism that advances the prior work in terms of search efficiency, search space coverage, and optimization metrics. NavCim introduces predictive modeling of ACiM hardware performance and uses the PPA prediction models instead of running simulators, significantly reducing search overheads. Faster searches enable NavCim to extend the architecture and model search spaces with an evolutionary search process to optimize architectures with more than two different tile sizes for multiple input models. With accuracy-aware searches, NavCim considers PPA and model accuracy together as optimization goals to achieve more balanced trade-offs. The experimental searches show that NavCim leverages predictive models to reduce search time by up to 7.3x without compromising the quality of search results. It also successfully identifies heterogeneous ACiM architectures that can efficiently execute multiple models on a single chip, improving accuracy by up to 19% over the prior work.Keywords
Funding Information
- National Research Foundation of Korea (RS-2023-00283799,A0457-20240100,RS-2023-00222663)
- Institute for Information and Communications Technology Promotion (2019-0-01906, 2021-0-00871, 2021-0-00310, RS-2023-00228970)
This publication has 43 references indexed in Scilit:
- ANN-Based Performance Prediction in MoCs2022
- Neuro-inspired computing chipsNature Electronics, 2020
- An investigation of latency prediction for NoC-based communication architectures using machine learning techniquesThe Journal of Supercomputing, 2019
- Machine Learning Based Framework to Predict Performance Evaluation of On-Chip Networks2018
- MobileNetV2: Inverted Residuals and Linear Bottlenecks2018
- PipeLayer: A Pipelined ReRAM-Based Accelerator for Deep Learning2017
- Deep Residual Learning for Image Recognition2016
- ImageNet: A large-scale hierarchical image database2009
- A Calculator for Pareto Points2007
- Stacked generalizationNeural Networks, 1992