運用頻譜分析法精算程序執行段落

A Precise Program Phase Identification Method Based on Frequency Domain Analysis

林軒毅(Hsuan-Yi Lin)

指導教授 ： 蔡仁松

國立清華大學/電機資訊學院/資訊工程學系所/碩士(2017年)

摘要

在本論文中，我們提出了一套系統化的方法，將程式執行的行為蹤跡轉換成以頻域表示，並藉此準確的定義程式執行段落。分析過的結果可以被嵌入程式碼中來標記每一個執行段落的起始點以及特性(例如CPI值)，並可以用來在程式運行時預測執行段落。藉由這些精確的程式執行段落資訊，將可以探索和開發更聰明的軟體和系統優化技術。

關鍵字

程序執行段落 ； 頻域分析 ； 基本區塊

並列摘要

In this paper, we present a systematic approach that transforms the program execution trace into frequency domain and precisely identifies program phases. The analyzed results can be embedded into program code to mark the starting point and execution characteristics, such as CPI (Cycles per Instruction), of each phase. The so generated information can be applied to runtime program phase prediction. With the precise program phase information, more intelligent software and system optimization techniques can be further explored and developed.

並列關鍵字

program phase ； frequency domain analysis ； basic block

參考文獻

[2] N. Peleg, B. Mendelson, “Detecting change in program behavior for adaptive optimization,” Proceedings of 16th International Conference on Parallel Architectures and Compilation Techniques (PACT), IEEE, 2007.

[3] R. Sarikaya, C. Isci, A. Buyuktosunoglu, “Runtime workload behavior prediction using statistical metric modeling with application to dynamic power management,” Proceedings of 2010 IEEE International Symposium on Workload Characterization (IISWC).

[4] L. Sawalha, S. Wolff, Monte P. Tull, Ronald D. Barnes, “Phase-guided scheduling on single-ISA heterogeneous multicore processors,” Proceedings of 14th Euromicro Conference on Digital System Design (DSD), IEEE, 2011.

[6] S. Padmanabha, A. Lukefahr, R. Das, S. Mahlke, “Trace based phase prediction for tightly-coupled heterogeneous cores,” Proceedings of the 46th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO), ACM, 2013.

[7] J. Lau, J. Sampson, E. Perelman, G. Hamerly, B. Calder, “The Strong correlation Between Code Signatures and Performance,” Proceedings of 2005 International Symposium on Performance Analysis of Systems and software (ISPASS), IEEE, 2005.