Back to Search Start Over

On the Impact of Subnet Clustering in Radio Hub for 100-core Wireless Network-on-Chip Architecture.

Authors :
Lit, Asrani
Sapawi, Rohana
Kipli, Kuryati
Suhaili, Shamsiah
Mahyan, Fariza
Fasiku, Ayojedi
Source :
International Journal of Computing & Digital Systems; Feb2024, Vol. 15 Issue 1, p1149-1160, 12p
Publication Year :
2024

Abstract

The integration of numerous embedded cores onto a single die is made possible by the Network-on-chip (NoC), which serves as a crucial technology. The utilization of planar metal interconnects for implementing a NoC through existing methods is inadequate because of the use of multihop channels in data exchange leads to significant power consumption and increased latency. Scalability challenges are anticipated to affect communication systems within many-core architectures expected in the future. As a solution, the Wireless Network-on-Chip or WiNoC design paradigm has been proposed to address these issues. In this paper, we evaluate how subnet clustering in the radio hub affects the 100-core Wireless Network-on-Chip architecture. This study investigates the overall delay in data transmission, the throughput of the network and energy usage in the 100-core WiNoC architecture with equipped with four radio hubs by analyzing single, four, and nine-tile radio subnet clustering. The validation of the results involves simulating the 100-core mesh WiNoC architectures being tested using the cycle-accurate Noxim simulator, with both random and transpose synthetic traffic workloads. Based on the simulation results, the architecture with a four-tile radio hub subnet clustering provides the most optimal performance for both random and transpose traffic distributions at PIR 0.013 flit/cycle/tile when compared to the other analyzed subnet clusterings. [ABSTRACT FROM AUTHOR]

Details

Language :
English
ISSN :
25359886
Volume :
15
Issue :
1
Database :
Complementary Index
Journal :
International Journal of Computing & Digital Systems
Publication Type :
Academic Journal
Accession number :
176160182
Full Text :
https://doi.org/10.12785/ijcds/150181