Identifying Program Power Phase Behavior using Power Vectors

Characterizing program behavior carries significant value in various avenues of computer engineering research from investigation of future architectures tailored better for emerging applications to OS based dynamic management techniques. Most modern applications exhibit distinctively different behavior throughout their runtimes, which constitute several phases of execution that share a greater amount of resemblance within themselves compared to other regions of execution. These execution phases can occur at scales comparable to total program execution, necessitating prohibitively long simulation times for characterization. Due to the implementation of extensive clock gating and additional power and thermal management techniques in modern processors, these program phases are also reflected in program power behavior, which can be used as an alternative means of program behavior characterization for power-oriented research. In this paper we present our methodology for identifying phases in program power behavior and determining execution points that correspond to these phases as well as defining a small set of power signatures representative of overall program power behavior. In our similarity analysis we use power vectors sampled at program runtime with our power estimation setup, which consist of power values for 22 processor sub-components. We define a power similarity metric as an intersection of both magnitude based and ratio-wise similarities and develop a thresholding algorithm in order to partition the power behavior into similarity groups. We illustrate our methodology with the Gzip benchmark for its whole runtime and characterize Gzip power behavior with both the selected execution points and defined signature vectors.