A View Of The Famous Castle


Published on

Published in: Technology
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide
  • A View Of The Famous Castle

    1. 1. Programmable Architectures for Communication Systems D. K. Arvind Institute for Computing Systems Architecture, Division of Informatics, The University of Edinburgh, Mayfield Road, Edinburgh EH9 3JZ, Scotland. Email: dka@dcs.ed.ac.uk
    2. 2. Edinburgh - The Capital City
    3. 3. Overview <ul><li>University of Edinburgh </li></ul><ul><li>Division of Informatics </li></ul><ul><li>Edinburgh InfoLab </li></ul><ul><ul><li>Research </li></ul></ul><ul><ul><li>Collaboration </li></ul></ul>
    4. 4. University of Edinburgh <ul><li>Founded in 1583 </li></ul><ul><li>Student Population - 18,023 : </li></ul><ul><ul><li>Undergraduate - 15,350; Postgraduate - 2,673 </li></ul></ul><ul><li>Staff - 6,649 : </li></ul><ul><ul><li>Academic Staff - 3,312 </li></ul></ul>
    5. 5. “ Informatics is the study of the structure, behaviour, and interactions of both natural and artificial computational systems.” ( http://www.informatics.ed.ac.uk/) Division of Informatics <ul><li>Institute for Adaptive and Neural Computation </li></ul><ul><li>Centre for Intelligent Systems and their Applications </li></ul><ul><li>Institute for Communicating and Collaborative Systems </li></ul><ul><li>Institute for Perception, Action and Behaviour </li></ul><ul><li>Laboratory for Foundations of Computer Science </li></ul><ul><li>Institute for Computing Systems Architecture </li></ul>
    6. 6. Division of Informatics <ul><li>Informatics@Edinburgh enjoys an international reputation for both its teaching and research </li></ul><ul><ul><ul><li>Only department in the UK awarded the top 5*A research rating in Computer Science in Dec. 2001 </li></ul></ul></ul><ul><ul><ul><li>UK’s biggest department with 87 research-active staff and 165 PhD students </li></ul></ul></ul><ul><ul><ul><li>Edinburgh-Stanford strategic research partnership </li></ul></ul></ul><ul><ul><ul><li>Location of the National e-Science Centre </li></ul></ul></ul><ul><li>Awarded top Excellent rating in the Teaching Quality Assessment </li></ul>
    7. 7. The Future …. <ul><li>Proliferation of Peer-to-Peer computing </li></ul><ul><ul><li>fundamental force of change and restructuring </li></ul></ul><ul><ul><li>Examples </li></ul></ul><ul><ul><ul><li>Cybiko - P2P wireless networked games </li></ul></ul></ul><ul><ul><ul><li>Napster - P2P sharing of music </li></ul></ul></ul><ul><ul><ul><li>Freenet - P2P information store </li></ul></ul></ul><ul><ul><ul><li>DoCoMo – P2P communication </li></ul></ul></ul><ul><li>Unregulated communications channels </li></ul><ul><ul><li>ISM, UWB, free-space optics, ….. </li></ul></ul><ul><li>System-on-Chip components </li></ul><ul><ul><li>banalisation of silicon technology </li></ul></ul><ul><li>Silicon falling behind </li></ul><ul><ul><li>storage & bandwidth improving at a greater rate </li></ul></ul>
    8. 8. P2P systems - Challenges <ul><li>Portability - Java, .NET </li></ul><ul><li>Performance - exploit concurrency </li></ul><ul><li>Mobility - size and energy consumption </li></ul><ul><li>Flexibility - soft- and hard-programmability </li></ul>
    9. 9. Research Focus <ul><li>“ To explore novel architectures for P2P systems using banalised technology, and enlighten future development of disruptive products and business change” </li></ul><ul><li>Our research is seeking programmable solutions which : </li></ul><ul><ul><li>harness progress in (a) technology (b) theory </li></ul></ul><ul><ul><li>implement high-performance algorithms and applications efficiently </li></ul></ul>
    10. 10. Disruptive technology opportunities <ul><li>System Architectures to explore … </li></ul><ul><li>Personal switch/P2P processor </li></ul><ul><li>Hubless, ephemeral, transient networks </li></ul><ul><li>Info-torch/Info-Klieg light </li></ul><ul><li>P2P n library|phone|gaming </li></ul>
    11. 11. Trends in the silicon fabric <ul><li>Convergence of transduction, communication and computation - heterogeneous systems with sensors and actuators </li></ul><ul><li>High performance computation at modest power consumption </li></ul><ul><li>Pre-designed IP blocks with different timing characteristics </li></ul><ul><li>The dominance of programmable fabrics - both soft- and hard-programmable </li></ul><ul><li>The complexity of the designs will demand novel architectures and design styles </li></ul>
    12. 12. The Die Area reachable in 1 clock cycle (1.2 GHz) At 0.1um (1 Billion transistors) only 16% of the chip is reachable in 1 clock cycle Dominance of interconnect delays over computational ones Network of Temporal Regions
    13. 13. Micronets - An alternative vision of Systems Architecture Micronet or Network-on-Chip : a network of entities which operate concurrently and communicate asynchronously Fractal model of system design: network of sub-systems, down to network of transistors Control is layered and distributed locally - behaviour can be decomposed to run on architectural clusters with the optimal mix of computational elements A clean separation between computation and communication, and, behaviour and timing - leads to a compositional design style
    14. 14. Behaviour-Architecture Co-design <ul><li>Integration Platforms composed of networks (micronets) of heterogeneous computational entities that operate in a multi-threaded fashion. </li></ul><ul><li>Applications composed of software blocks: some pre-defined, such as communication protocols; others, more specific to the application. </li></ul><ul><li>Co-design (Step 1) : recognise concurrent operations and optimise communication at different levels of granularity in the application and map them to the platform </li></ul><ul><li>Co-design (Step 2) : explore the trade-off between programmability (both soft- and hard-), and performance (MOPS/mWatt) of the application running on the platform </li></ul>
    15. 15. The COMPASS Design Environment <ul><li>Visualisation of energy and performance effects of compiler optimisations </li></ul><ul><li>Distributed simulation platform on a 16-node Beowulf cluster </li></ul><ul><li>Java or C applications </li></ul><ul><li>SSA intermediate representation </li></ul><ul><li>Soft- and Hard-programmability </li></ul>
    16. 16. Design framework for programmable multi-threaded systems
    17. 17. A micronet-based multi-threaded architecture
    18. 18. void Micronet(chan tinst Inst, chan tpc Pc, chan tregval RegDump, chan Word MemDump) { //Define channels chan tinst ALUinst, MUinst; chan tpc ALUpc; chan tack ALUCUack, MUCUack; chan tregreq RegRequest; chan tregreturn Xout, Yout; chan tregval ALUXin, ALUYin, MUXin, MUYin; chan twriteback toReg, ALUWBout, MUWBout; chan bool KillBus; //Spawn linked Functional Units in Parallel //+ clock 32 par{ //Buffers for register requests //+ clock 32 ControlUnit(Inst, Pc, ALUinst, MUinst, ALUCUack, MUCUack, ALUpc, RegRequest, KillBus ); //+ clock 50 RegisterBank(RegRequest, //Requests //Lock writeback registers Xout, Yout, //To the bus toReg, RegDump); //Writeback //X bus //+ clock 32 BusSplit(Xout, ALUXin, MUXin); //Y bus //+ clock 32 BusSplit(Yout, ALUYin, MUYin); //+ clock 50 ALU(ALUinst, ALUXin, ALUYin, ALUCUack, ALUpc, ALUWBout); //+ clock 32 MU(MUinst, MUXin, MUYin, MUCUack, MUWBout, MemDump); //Writeback bus //+ clock 32 BusMerge(ALUWBout, MUWBout, toReg, KillBus); } Automatic Synthesis of Micronet Architecture from Specification
    19. 19. Power/Speed estimations on the M/T architecture Overall TPU 0 TPU 1
    20. 20. Power - Speed Tradeoff for Programs executing on Micronet Architectures
    21. 21. Example of an Internet Appliance <ul><li>Bluetooth-based system in VCC </li></ul><ul><li>Two physical objects: the WAPmobile, and a WAP `phone </li></ul><ul><li>The behaviour of an internet- and Bluetooth-enabled Basestation, and a Bluetooth-enabled robot is simulated in VCC </li></ul><ul><li>The WAP phone controls the robot in real-time via the VCC behavioural models </li></ul>
    22. 22. Proven Research Expertise in Systems Architecture <ul><li>Programmable Architecture Design </li></ul><ul><ul><li>Micronet-based asynchronous architectures </li></ul></ul><ul><ul><li>Java and C compilation for multi-threaded embedded systems </li></ul></ul><ul><ul><li>Applications include Bluetooth- and 802.11-based ones </li></ul></ul><ul><li>Vertically-integrated environment (COMPASS) for energy-conscious, high-performance embedded system design </li></ul><ul><li>Industrial research partners </li></ul><ul><li>Well-endowed laboratory, including a 16-node Beowulf cluster for simulations and state-of-the-art EDA tools </li></ul>
    23. 23. Model for Collaboration <ul><li>Feature set </li></ul><ul><ul><li>‘ Beyond the envelope’ research </li></ul></ul><ul><ul><li>Pre-competitive : several industrial partners </li></ul></ul><ul><ul><ul><li>industrial support : funding, equipment, body swap,…. </li></ul></ul></ul><ul><li>Successful Examples </li></ul><ul><ul><li>Silicon Structures (Caltech 1977 - 81) </li></ul></ul><ul><ul><li>Berkeley Wireless Research Center (1998 - ) </li></ul></ul><ul><ul><li>MIT Media Lab (1987 - ) </li></ul></ul>
    24. 24. Road Map <ul><li>Creation of the “ Edinburgh InfoLab ” to research architectures for future P2P systems </li></ul><ul><li>5 founding industrial partners/subscribers </li></ul><ul><li>30 PhD students in the steady state </li></ul><ul><li>Partners’ contributions: Two 4-year PhD studentships per year, cumulatively for 3 years </li></ul><ul><li>Interested? Email: dka@dcs.ed.ac.uk </li></ul><ul><li>More details at http://www.dcs.ed.ac.uk/~dka </li></ul>
    25. 25. Thank You