Abstract

On-chip implementation of multiprocessor systems needs to planarise the interconnect networks onto the silicon floorplan. Compared with traditional ASIC/SoC architectures, Multiprocessor Systems on Chips (MPSoC) node processors are homogeneous, and MPSoC network topologies are regular. Therefore, traditional ASIC floorplanning methodologies that perform macro placement are not suitable for MPSoC designs. We propose an automated MPSoC physical planning methodology. REGULAY can generate an optimal floorplan for different topologies under different design constraints. Compared with traditional floorplanning approaches, REGULAY shows significant advantages in reducing the total interconnect wirelength while preserving the regularity and hierarchy of the network topology.

Highlights

  • Future VLSI technology will enable hundreds, or even thousands, of Processing Elements (PEs) being integrated on the same chip

  • We propose a floorplanning method and a tool called REGULAY that can automatically place regularly-configured Multiprocessor Systems on Chips (MPSoC) node processors as well as switch fabrics onto a user-defined tile floorplan

  • We propose a two-step heuristic approach that takes into account the special properties of MPSoC topologies

Read more

Summary

Introduction

Future VLSI technology will enable hundreds, or even thousands, of Processing Elements (PEs) being integrated on the same chip. Similar to parallel computer clusters, homogeneous MPSoCs may adopt different network topologies and switch fabrics and implement them onto the chip floorplan. Prior network graph planarisation approaches either targeted only some specific topologies, or they were not flexible enough to adapt to many of the floorplan constraints imposed by the silicon implementation (Dehon, 2000; Greenberg and Leiserson, 1998). Those approaches are not suitable for an automated design flow. MPSoC planar layout requires that PE blocks are tiled and abutted on the floorplan in a two-dimensional tile array

MPSoC network topologies
Orthogonal topology
Cube-Connected-Cycles topology
Fully-connected network
Fat-tree topology
Butterfly topology
MPSoC network floorplan
Problem formulation
Regularity extraction
Forming the objective function
X-dimension optimisation
Y-dimension optimisation
Legalisation
Conclusion
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call