Performance Optimization of Cu Wires for Network-on-chip Based Many-core Architectures

Please enable javascript in your browser.

Radfar, Farzad | 2013

929 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 44719 (05)
University: Sharif University of Technology
Department: Electrical Engineering
Advisor(s): Sarvari, Reza
Abstract:
The exponential increase in power density within a chip due to higher frequency of operation in recent years (Moor's law) is a major limiting factor for designers. Increasing the number of parallel cores instead of increasing the frequency of operation is a possible solution. The design of connections within the cores can be followed by the old process but the global interconnectsbetween the cores instead of point to point can be replaced byNetwork-on-Chip (NoC). In this thesis, The dimensions of global interconnects in many-core chips are optimized for maximum bandwidth density and minimum delay taking into account network-on-chip router latency and size effects of coppe. The optimal dimensions thus obtained are used to characterize different network-on-chip topologies based on wiring area utilization, maximum core-to-core channel width, aggregate chip bandwidth and worse case latency. Finally, the advantages and disadvantages of many-core chips are evaluated for different network-on-chip topologies by considering the optimal design of global interconnects
Keywords:
Network-on-Chip (NOC) ; Router ; Delay Estimation Method ; Global Interconnect

No TOC