CAREER: System-Level Power/Performance Analysis for Embedded Systems Design

本研究的目标是提供一套正式和半正式的技术，系统级的功耗/性能分析，可以帮助设计人员选择合适的平台，从一组目标应用程序。平台是指一系列异构体系结构（由可编程和专用组件组成），这些体系结构满足一组体系结构约束，允许重用硬件和软件组件。遵循设计过程中关注点正交化的原则，这种方法基于明确区分应用程序和体系结构的关键思想。我们的系统级分析方法有三个主要目标：1）应用程序和平台建模2）性能模型评估3）性能模型验证。虽然所提出的方法是完全通用的，我们最初的重点是便携式嵌入式多媒体系统（例如，小型主机，如PDA、网络计算机）。对于这些系统，与安全关键应用中使用的反应式嵌入式系统相反，平均行为远比最坏情况下的行为更重要。所提出的方法补充了现有的极端情况下的性能分析技术，将环境特性纳入系统性能评估。针对系统级，这种分析的结果并不局限于任何特定的硬件/软件实现。综上所述，我们提出的技术将允许媒体系统设计人员更快地探索架构，更稳健地估计不同设计选择的影响，并更有效地使用大型多媒体数据基准。

The goal of this research is to provide a set of formal and semiformal techniques for system-level power/performance analysis that can help the designer to select the right platform starting from a set of target applications. By platform we mean a family of heterogeneous architectures (which consist of both programmable and dedicated components) that satisfy a set of architectural constraints imposed to allow re-use of hardware and software components. Following the principle of orthogonalization of concerns during the design process, this methodology is based on the key idea of making a clear distinction between applications and architectures. Our approach to system-level analysis has three main objectives: 1) Application and platform modeling 2) Performance model evaluation 3) Performance model validation. Although the proposed methodology is completely general, our initial focus is on portable embedded multimedia systems (e.g., slim hosts like PDAs, network computers). For these systems, as opposed to the reactive embedded systems used in safety critical applications, the average behavior is far more important than the worst-case behavior. The proposed methodology complements the existing techniques for extreme-case performance analysis by incorporating the environment characteristics into system performance evaluation. Being targeted at system-level, the results of this analysis are not confined to any particular hardware/software implementation. Taken together, our proposed techniques will allow media systems designers to explore architectures more rapidly, estimate the impact of different design choices more robustly, and use large multimedia data benchmarks more effectively.