A Loss-Aware Growing Ring Self-Organizing Map (GRSOM)-based Mapping Algorithm in Optical Network-on-Chip (ONoC)

Optical Network-on-Chip (ONoC) is promising to provide a high-performance on-chip communication for future many-core processors, with high bandwidth and low power consumption. However, the reliability challenges, due to the loss and crosstalk, limit the power efficiency and scalability of ONoC. This...

Full description

Saved in:
Bibliographic Details
Main Authors: Niu, Yuxiang, Li, Hui, Liu, Feiyang
Format: Conference Proceeding
Language:English
Subjects:
Integrated optics
loss
mapping algorithm
Network-on-chip
neural network
Optical losses
Optical Network-on-Chip (ONoC)
Power demand
Program processors
reliability
Scalability
Self-organizing feature maps
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Optical Network-on-Chip (ONoC) is promising to provide a high-performance on-chip communication for future many-core processors, with high bandwidth and low power consumption. However, the reliability challenges, due to the loss and crosstalk, limit the power efficiency and scalability of ONoC. This work proposes a Growing Ring Self-Organizing Map (GRSOM)-based mapping algorithm in O NoC, which optimizes the loss of on-chip optical communications, thus improving the power efficiency and reliability. The evaluation results show that the proposed mapping algorithm has advantages in terms of computational complexity, running speed and accuracy, and can meet the loss requirement.
ISSN:2576-7828
DOI:10.1109/ICCT56141.2022.10073320