Tinuso: A processor architecture for a multi-core hardware simulation platform

Pascal Schleuniger, Sven Karlsson

    Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearch

    1 Downloads (Pure)

    Abstract

    Multi-core systems have the potential to improve performance, energy and cost properties of embedded systems but also require new design methods and tools to take advantage of the new architectures. Due to the limited accuracy and performance of pure software simulators, we are working on a cycle accurate hardware simulation platform. We have developed the Tinuso processor architecture for this platform. Tinuso is a processor architecture optimized for FPGA implementation. The instruction set makes use of predicated instructions and supports C/C++ and assembly language programming. It is designed to be easy extendable to maintain the exibility required for the research on multi-core systems. Tinuso contains a co-processor interface to connect to a network interface. This interface allow for communication over an on-chip network. A clock frequency estimation study on a deeply pipelined Tinuso implementation shows that it is feasible to operate it at a frequency higher than 300 MHz in current state-of-the-art high end FPGAs.
    Original languageEnglish
    Title of host publicationProceeding of third swedish workshop on multi-core computing - MCC'10
    Volume3
    Publication statusAccepted/In press - 2010
    EventSwedish workshop on multi-core computing - Gothenburg, Sweden
    Duration: 1 Jan 2010 → …
    Conference number: 3

    Conference

    ConferenceSwedish workshop on multi-core computing
    Number3
    CityGothenburg, Sweden
    Period01/01/2010 → …

    Fingerprint Dive into the research topics of 'Tinuso: A processor architecture for a multi-core hardware simulation platform'. Together they form a unique fingerprint.

    Cite this