Deterministic parallel processing

  • Authors:
  • Gajinder Panesar;Daniel Towner;Andrew Duller;Alan Gray;Will Robbins

  • Affiliations:
  • picoChip Designs Limited, Bath, UK;picoChip Designs Limited, Bath, UK;picoChip Designs Limited, Bath, UK;picoChip Designs Limited, Bath, UK;picoChip Designs Limited, Bath, UK

  • Venue:
  • International Journal of Parallel Programming
  • Year:
  • 2006

Quantified Score

Hi-index 0.01

Visualization

Abstract

In order to address the problems faced in the wireless communications domain, picoChip has devised the picoArrayTM. The picoArray is a tiled-processor architecture, containing several hundred heterogeneous processors, connected through a novel, compile-time scheduled interconnect. This architecture does not suffer from many of the problems faced by conventional general purpose parallel processors and provides an alternative to creating an ASIC. The PC102 is the second generation device from picoChip containing 308 processors. The devices are designed to be connected together using a seamless extension of the internal interconnect structure. This enables multi-chip solutions to be easily realised for applications which require additional processing. This paper highlights some of the difficulties encountered when building parallel systems and goes on to show how the features of the picoArray allow deterministic processing to be achieved, how the tool chain allows programming to be performed effectively in a combination of high level assembly language and C, and how systems built around the picoArray are debugged in real-time. By handling a wide variety of types of processing within the picoArray a single design flow can be used to produce complex communications systems. The effectiveness of this approach is demonstrated through the use of the picoArray to build a 802.16 base-station for commercial deployment.