A High-Performance Scalable Computing System on the RapidIO Interconnect Architecture

Abstract

In order to meet the demanding requirements of calability, adaptability and computational capability of next-generation signal-processing system, a flexible and high performance signal-processing odule based on Serial RapidIO (SRIO) interconnect and Advanced Mezzanine Card (AMC) modules is developed. RapidIO is a packet-switched nterconnect intended primarily as an intra-system interface, allowing chip-to-chip and board-to-board communications at 10 Gigabit per second performance levels. The AMC specification is defined by PCI Industrial Computer Manufacturer Group (PICMG) and mainly covers the base-level requirements for a wide-range of high-speed mezzanine cards optimized for, but not limited to, Advanced Telecommunications Computing Architecture (ATCA) Carriers systems. With this versatile module, massively parallel multi-processor computing system is feasible. A prototype of 12 modules is introduced and an application of this computing system in MIMO-OFDM system is covered.