結合功率管理與運算排序之高階合成系統

運算排序在高階合成中是一項最為重要的工作，大部分傳統的運算排序演算法都是針對控制步驟最小化或運算資源最小化。然而隨著設計複雜度的提升，巨大的峰值功率可能會造成壓降效應而導致邏輯錯誤或是電子遷移效應而導致電路可靠度上的問題。又因無線通訊以及個人可攜式產品的普及，低功率已經成為重要的設計目標，降低平均功率將可以延長電池的使用時間以及節省能源的消耗。一種有效降低峰值功率以及平均功率的功率管理方法是使用控制邏輯將已知不需要執行的運算停止執行。在本篇論文中我們對降低峰值功率以及平均功率分別提出了整數線性規劃的式子來描述這個結合運算排序以及功率管理的問題。由整數線性規劃的式子我們可以得到關於這個問題的最佳解，但所需花費的時間較長。因此，我們又提出了一個啟發式演算法能在較快的時間內得到近似最佳解。由實驗結果顯示我們的方法確實能有效的降峰值攻率以及平均功率。

關鍵字

功率管理；啟發式演算法；平均功率；電子遷移效應；運算排序；整數線性規劃；控制步驟；峰值功率；壓降效應；運算資源

並列摘要

Operation scheduling is the most important task in highlevel synthesis. Most conventional operation scheduling algorithms paid attention to the minimization of control steps or the minimization of resources. However, as the design complexity continues to increase, huge peak power may lead to logic errors due to voltage drops or reliability problems due to electromigration. Due to the popularization of portable and wireless communication, low power design is getting very important, to reduce average power can improve the life of battery and the energy saving. An efficient power management method is to use control logic to shut down unused operations. In this paper, we present an ILP (integer linear programming) formulation to model the problem of combining operation scheduling and power management for peak power reduction and average power but the run time is too long. Therefore, we propose a heuristic algorithm achieves the near optimal solution. Benchmark data show that our approach can reduce peak power and average power effectively.

並列關鍵字

heuristic algorithm ； electromigration ； power management ； average power ； ILP ； control steps ； operation scheduling ； resources ； peak power ； voltage drop

參考文獻

[5] G. De Micheli, “Synthesis and Optimization of Digital Circuits”, McGraw-Hill Academic Publishers, 1994.

[6] R. Jain, A. Mujumdar, A. Sharma and H.M. Wang, “Empirical Evaluation of Some High-Level Synthesis Scheduling Heuristics”, Proc. of IEEE/ACM Design Automation Conference, pp. 686—689, 1991.

[8] P. Faraboschi, J.A. Fisher and C. Young, “Instruction Scheduling for Instruction Level Parallel Processors”, Proc. of the IEEE, vol. 89, no. 11, pp. 1638—1659, 2001.

[9] S. Davidson, D. Landslov, B.D. Shriver and P.W. Mallett, “Some Experiments in Local Microcode Compaction for Horizontal Machines”, IEEE Trans. on Computers, pp 460—477, 1981.

[12] S. H. Huang, C.T. Hwang, Y.C. Hsu, and Y. J. Oyang, “A New Approach to Schedule Operations across Nested-ifs and Nested-loops”, Proc. of IEEE/ACM International Symposium on Micro-architecture, pp. 268—271, 1992.

國際替代計量

結合功率管理與運算排序之高階合成系統

未授權

主題瀏覽