II-EN: Fast Thermal-Aware Sampling for Multiprocessors

该提案的目的是调整计算机架构仿真基础设施，以模拟未来多核系统的功耗和散热问题。该建议解决的主要问题是当还考虑功率和热估计时的仿真速度。为了加快模拟时间，工程师们采用了采样。 问题是，添加现代多核和热模拟会破坏现有的采样技术。 研究界需要性能、功耗和热仿真结果至少比当前可用的速度快一个数量级。这一建议为建立这样的系统提供了支持。仿真基础设施将对计算机体系结构社区产生重大影响。我们预计第三方出版物的数量会增加。我们还希望更大的，多大学，类课程的影响，由于使这个模拟器可用于类项目。结合前两个影响，我们预计更大比例的研究生和本科生将学习和研究未来的多核、移动的系统和云，重点关注电源和散热问题。

The objective of this proposal is to adapt computer architecture simulation infrastructure to model power and thermal issues for future multicore systems. The main concern addressed by this proposal is simulation speed when power and thermal estimations are also considered. To accelerate simulation times, engineers employ sampling. The problem is that adding modern multicores and thermal simulation breaks existing sampling techniques. The research community needs performance, power, and thermal simulation results at least one order of magnitude faster than currently available. This proposal provides support to build such system.The simulation infrastructure will have a significant impact on the computer architecture community. We expect an increase in the number of publications by third parties. We also expect a larger, multi-university, class curriculum impact due to make this simulator available to class projects. Combining the two previous impacts, we expect a larger proportion of the graduate and undergraduate students to learn and research about future multicores, mobile systems, and clouds with a focus on power and thermal issues.