The University of Florida 
High-performance Computing & Simulation Research Lab
home > project > hpccgroup

    submenu »     project home | overview | downloads | publications | related links

HPCC Group / Project Overview
Scalable and Dependable Applications and Infrastructure for High-Performance Computing and Networking

Reconfigurable Hardware Empowered Grid Computing

MOTIVATIONS

      *   Building run-anywhere applications seen as cost-effective approach for future HPC applications

*   Grid computing tries to solve the problem of not enough resources for large HPC applications

*   Customized hardware try to solve the problem of not enough speed for large applications

*   RC (FPGA-based) provides flexibility over regular hardware (ASIC-based) computing and solves the problem of I am stuck to the same hardware!

*   So why not RC-enabled Grids: flexible, fast computation with shareable resources spread across sites that are physically apart from each other

 

APPROACH

      *  Study the challenges involved in enabling Grids with RC using simulation

*  Application performance feasibility study on future RC grids

* Extensions to Grid middleware such as Globus and other grid tools

*  Initial deployments to assess validity of assumptions

*  Will leverage a strong background in Grid computing and RC

~V        International Virtual Data Grid Laboratory (iVDGL) -- NSF

~V        MONitoring Agents using a Large Integrated Services Architecture (MonALISA) -- CalTech

~V        Gossip-Enabled Monitoring Service (GEMS) -- UF

~V        SPHINX: Grid Scheduling Middleware -- UF

~V        Comprehensive-based Approach to Reconfigurable Management Architecture (CARMA) -- UF

 

CHALLENGES (Foreseen for RC Grids)

  *   VHDL, configuration files and the like are more architecture-dependent than regular C or Fortran routines

    *    Locating, transferring and handling of I/O and configuration files over long haul networks

    *    Timing issues w.r.to preloading the configuration files to program the FPGAs

         *    Handling dynamic reconfiguration in real time

    *   Partitioning of large applications into multiple FPGAs

    *   Additionally, general challenges related to Grids

 

 

 

OTHER HPCC Group PROJECTS

GEMS Project

Parallel and Distributed Computing for Fault-tolerant Sonar Arrays

Computational Framework for Simulating Joint Mechanics

High-Performance Networking Related Projects