Parallel The Development Of Multi-core Technology, Test And Measurement Applications

Single-core processors and the performance imperative of Moore's Law is approaching an upper limit in terms of adding increasing processing power simply by increasing clock speeds.In response to this problem,processor suppliers have turned to multi-core architectures in order to achieve performance gains.From currently dual-,quad-,octo-core chips to possible 80 cores in 2011,the computing world has inevitably entered the multi-core age. The technology advancement of microprocessor brings great opportunity to PC based test and measurement applications.However,to take full advantage of multicore processors,engineers must develop parallel,multithreading test and measurement applications.Amdahl's law tells us that the speedup of a program using multi-core processors in parallel computing depends on the parallel fraction of the program,so the most challenging part is the area of software development.This thesis is going to propose a brand new design methodology for multi-core: Patterns for Parallel Programming based on Graphical Platform LabVIEW, which could be effectively applied to achieve the performance increase with multi-core.Benefiting from the inherent parallelism and automatic multithreading management in LabVIEW,the thesis puts more emphasis on the design and analysis methodology like how to find the concurrency and design the algorithm structure.Specifically for test and measurement applications,three parallel programming strategies are proposed as Task Parallelism,Data Parallelism and Pipelining.To introduce these strategies,it starts from how to think parallel,analyze patterns based on application characteristics,to best practices in LabVIEW,as well as the benchmark experiment and industrial application analysis.This is a brand new idea for parallel programming for multi-core,which should also be a good reference architecture for test and measurement engineers and developers.Finally,we are looking forward to more innovative application areas opened with multi-core technology,like Symmetric Multiprocessing(SMP) in Real-Time Operating System,up to 1M FFTs/sec computing capability and virtualization technology on multi-core PCs etc.