SMA: Accurate Temperature Measurement Infrastructure and Methodology for Power, Variability, and Reliability Analysis

在计算机系统的设计、原型制作、性能评估和分析中，仿真环境是不可或缺的工具。模拟器必须能够真实地反映被分析系统的行为。为了保证模拟器的准确性，必须对其进行检定，使其与化学数据紧密匹配。现代处理器提供了足够的性能计数器来验证大多数性能模型，但所提供的信息不足以验证功耗、热模型和工艺可变性。温度和功耗是现代高性能体系结构的首要设计参数。较高的工作温度和较大的功率消耗可能会限制性能和可制造性。虽然温度和功率是处理器设计的主要因素，但工艺可变性正在成为另一个主要限制因素。为了解决与功耗、热模型和工艺可变性验证相关的一些困难，我们建议使用红外测量设置来捕获现代芯片的运行时功耗和热特性。详细的热模型可以使用测量的温度来生成电源详细功耗。同样的热量测量将用于测量现成处理器的工艺可变性。此外，将使用硫化基础设施来开发经过验证的热模拟基础设施和工艺可变性模型。

Simulation environments are an indispensable tool in the design, prototyping,performance evaluation, and analysis of computer systems. Simulator must be ableto faithfully reflect the behavior of the system being analyzed. To ensure theaccuracy of the simulator, it must be verified and determined to closely matchempirical data. Modern processors provide enough performance counters tovalidate the majority of the performance models; nevertheless, the informationprovided is not enough to validate power consumption, thermal models, andprocess variability.Temperature and power consumption are first order design parameters for modernhigh performance architectures. High operational temperatures and large powerconsumption present possible limits to performance and manufacturability. Whiletemperature and power are dominant factors on processor design, processvariability is becoming another major constraint.In order to address some of the difficulties associated with the validation ofpower, thermal models, and process variability, we propose to use an infraredmeasurement setup to capture run-time power consumption and thermalcharacteristics of modern chips.A detailed thermal model can use the measured temperatures to generate a powerdetailed power consumption. The same thermal measurements will be used tomeasure process variability from of-the-shelf processors. In addition, theresulting infrastructure will be use to develop a verified thermal simulationinfrastructure and process variability models.