NGS: Collaborative Research: Continuous Compilation: A New Approach to Aggressive and Adaptive Code Transformation

今天优化研究面临的挑战是开发新的技术和方法，这些技术和方法可以产生早期优化研究中典型的20%到40%甚至更多的性能改进。在本研究中，我们通过调查和开发创新系统或应用优化来解决这一挑战。我们的系统，连续编译系统（CoCo），使用在编译时开发并在运行时调整的优化计划，在编译时静态地和在运行时动态地应用优化。为了演示我们方法的实用性和实用性，我们首先将CoCo应用于几个大型的、长时间运行的应用程序。此类应用的示例包括生态模拟、天气模拟、架构模拟器和VLSI路由和放置应用。随着我们对连续编译应用程序的深入了解和经验的积累，我们将研究它在其他类型的代码（如多媒体应用程序、信号处理应用程序和web服务器）上的有效性。

Jack W. Davidson Illinois Institute of Technology Collaborative Research: Continuous Compilation: A New Approach to Aggressive and Adaptive Code Transformation Today's challenge for optimization research is to develop new techniques and approaches that yield performance improvements that were typical of early optimization research-20 to 40 percent and more. In this research, we address this challenge by investigation and developing an innovative system or applying optimizations. Our system, the Continuous Compilation System (CoCo), applies optimizations both statically at compile-time and dynamically at run-time using optimization plans developed at compile time and adapted at run time. To demonstrate the practicality and utility of our approach, initially we will apply CoCo to several large, long-running applications. Examples of such applications include ecological simulations, weather simulation, architecture simulators, and VLSI routing and placement applications. As we gain insight and experience with the application of continuous compilation, we will investigate its effectiveness on other types of codes such as multimedia applications, signal processing applications, and web servers.