Contribution to a conference proceedings

FZJ-2024-00959

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png An efficient synchronous-updating memristor-based Ising solver for combinatorial optimization

Jiang, M. ; Shan, K. ; Sheng, X. ; Graves, C. ; Strachan, J. P. (Corresponding author)FZJ* ; Li, C.

2022
IEEE

This record in other databases:  

Please use a persistent id in citations: doi:10.1109/IEDM45625.2022.10019348

Abstract: Despite showing significant potential in solving combinatorial optimization problems, existing memristor-based solvers update node states asynchronously by performing matrix multiplication column-by-column, leaving the massive parallelism of the crossbar not fully exploited. In this work, we propose and experimentally demonstrate solving the optimization problems with a synchronous-updating memristor-based Ising solver, which is realized by a binary neural network-inspired updating algorithm and a physics-inspired annealing method. The newly proposed method saves more than 5x time and 35x energy consumption compared to the state-of-the-art mem-HNN for finding the optimal solution to a 60-node Max-cut problem.

---

Contributing Institute(s):

1. Neuromorphic Compute Nodes (PGI-14)

Research Program(s):

1. 5234 - Emerging NC Architectures (POF4-523) (POF4-523)
2. 5233 - Memristive Materials and Devices (POF4-523) (POF4-523)

---