F5-11: Device Performance Metrics and Mission-Critical Processing Dr. Alan D. George CHREC F5/F6 WEEKLY MEETING Dr. Herman Lam Nicholas Wulf Justin Richardson Quinn Martin Steven Fingulin Assoc. Professor of ECE University of Florida Research Students University of Florida Professor of ECE University of Florida Dr. Ann Gordon-Ross Asst. Professor of ECE University of Florida F5-11: Plans for Spring 2011 Justin Richardson CHREC F5/F6 WEEKLY MEETING CHREC F5: Justin Task A Metrics Tool (Justin and Quinn) Port existing application to a Webapp format for portability, ease of access, and expandability Expand application to include remaining device metrics and parameters Benchmarking (Justin, Seth, and Steven) RU Gather more academic and professional literature to allow a narrower focus Benchmark using expert tools (ie GPU/FPC comparison) Refine methodology and definitions of α for clarity New Platforms Finish GPU benchmarking Add ARM to benchmark suite and get basic benchmarking started Gantt Chart http://www.hcs.ufl.edu/~richardson/Projects/Gantt-Chart-SP2011.html 3 F5-11: Plans for Spring 2011 Steven Fingulin CHREC F5/F6 WEEKLY MEETING Spring Tasks Device Studies Add new devices to analysis Add XML entries to the tool for calculation of other metrics Compare devices based on complete metrics suite Achronix FPGAs (Stratix V, Virtex 7) New CPU/GPU hybrid chips (AMD Fusion, Intel Sandy Bridge, Tegra 2, etc.) Many-core devices (Tilera and similar) DSPs (TI OMAP and AD) New embedded processors (Snapdragon, Scorpion, Cortex A9, etc.) Low power families of major processors Measure and analyze power use Add new and larger benchmarks Gantt Chart http://hcs.ufl.edu/~fingulin/Spring2011research.png F5-11: Plans for Spring 2011 Quinn Martin CHREC F5/F6 WEEKLY MEETING CHREC F5: Quinn Task A Metrics Tool Port existing tool to web application Ajax integration with RU database Recode GUI for web Expand application Add Memory-based metrics (IMB, EMB) Add IOB Expand and document XML format Gantt Chart http://www.hcs.ufl.edu/~martin/gantt_chart_sp2011.png 7 F5-11: Plans for Spring 2011 Nick Wulf CHREC F5/F6 WEEKLY MEETING Task B: Mission-Critical Processing Processing challenges in restricted environments Complex optimization problem for given mission Small satellites, UAVs, smart munitions, AUVs Radiation effects; limited size, weight, & power (SWaP) Initial focus upon military/aerospace missions Goals (throughput, response time, dependability) Applications (sensor processing, autonomy, et al.) Constraints (power, size, mass, cooling, wiring) Architecture (fixed, reconfigurable, hybrid; device & subsystem) Reliability (RHBP, RHBD, NMR, SIFT, ABFT, et al.) Pursuit of novel optimization theory, data, tools Given mission goals, apps, & constraints, what is optimal existing or notional architecture (device, subsystem) to address? UAV = unmanned aerial vehicle AUV = autonomous underwater vehicle 9 RHBP/D = RadHard by process/design NMR = N-modular redundancy SIFT = software-implemented FT ABFT = algorithm-based FT