Power consumption and related thermal effects of processor chips are worstening with each technology generation. Modeling the power and thermal behavior of processor chips requires challenging and complex validation approaches which, in some cases, may be unreliable. The proposed infrastructure project develops an infrared measurement setup that allows run-time power consumption of state-of-the-art processors to be accurately measured in real time. The proposed infrared measurement setup has a high spatial resolution (10x10um) with a high frame rate (125Hz) capacity to capture thermal maps of processor chips. The synchronized traces produced can be used to validate existing and newly proposed power and thermal models, and they can also be used to develop new process variability models for the design of future generation processor chips.
Temperature is a dominant factor in the performance, reliability, and leakage power consumption of modern processors. As a result, increasing numbers of researchers evaluate thermal characteristics in their proposals. In this paper, we measure a real processor focusing on its thermal characterization executing diverse workloads. This NSF proposal has provided the infrastructure necessary to perform such measurements. The NSF CRI award has allowed the purchase of infrared equipment necessary to measure state of the art chips. The infraestructure has allowed to develop new and more efficient methods to measure unmodified chips. Previous Infrared Thermography infrastructures has been developed to measure the temperature in real-time. Since the infrared opaque metal heat sinks need to be replaced with an infrared transparent heat sink in these setups, oil based cooling solutions have been proposed. However, oil is not a representative of a metal heat sink because measurement with oil based cooling can change the thermal behavior of the processor. This work has allowed to build arepresentative oil based cooling solution, and show that it has the same thermal response as a metal heat sink. The resulting setup has been used to produce several publications, and more importantly, it has been used as a reference model for other academics and industry. Companies like NVIDIA and Oracle have used the setup and modified their internal setups to obtain better results. Beyond the current application in industry, the setup has allowed to develop a real time measurement with a thermal equivalent heat sink that did not exists. As a summary, this project has benefitted industry and academia with current and future problems.
