本文選題：多核處理器 + 進程調(diào)度　； 參考：《哈爾濱工程大學》2012年碩士論文

【摘要】：隨著多核技術(shù)的發(fā)展，多核處理器進程調(diào)度已成為目前高性能計算研究領(lǐng)域的熱點之一。近年來，針對多核處理器進程調(diào)度課題，國內(nèi)外諸多專家學者和科研機構(gòu)對其展開了積極的研究和探索，旨在通過進程快速分配、適應(yīng)性調(diào)整，以提高處理器的進程調(diào)度效率。但是，目前多核處理器的進程調(diào)度算法仍不夠理想，存在進程分配效率不高、動態(tài)調(diào)整算法復雜度高、負載不均衡現(xiàn)象嚴重等問題，嚴重影響了多核處理器的進程調(diào)度性能。因此，本文將針對多核處理器的進程調(diào)度算法展開研究，以尋求更為高效的進程調(diào)度算法。 本文在分析和研究現(xiàn)有多核處理器進程調(diào)度算法的基礎(chǔ)上，針對現(xiàn)有高效進程調(diào)度算法存在的不足，提出一種綜合性高效的進程調(diào)度算法（Adaptive Fast allocationscheduling，AFAS）。AFAS調(diào)度算法與已有算法相比，本算法的技術(shù)優(yōu)勢：通過結(jié)合區(qū)間插入技術(shù)和關(guān)鍵節(jié)點算法，降低處理器的空閑時間，確保進程執(zhí)行的有序性。采用環(huán)劃分算法對進程進行分塊處理，提高處理器的進程分配效率；根據(jù)進程的約束條件展開啟發(fā)式檢索，并利用回溯追蹤機制，確保進程分配方案的最優(yōu)性；利用處理器負載界限值調(diào)節(jié)進程的負載均衡性，減少進程遷移產(chǎn)生的系統(tǒng)資源開銷。同時，針對多核處理器的多進程并行分配，設(shè)計具體的分配方案，有利于將進程分配算法更好地應(yīng)用于實踐，提高多核處理器的進程分配性能。 為驗證綜合性高效進程調(diào)度算法的可行性和高效性，本文通過設(shè)計合理的性能測試方案，對其進行性能測試和結(jié)果分析。實驗驗證結(jié)果表明：新型的進程分配算法有效地克服了傳統(tǒng)算法存在的進程分配效率低下、負載均衡性不高等缺點，有效地提高了多核處理器的進程分配性能，具有良好的應(yīng)用前景和理論研究價值。
[Abstract]:With the development of multi-core technology, multi-core processor process scheduling has become one of the hotspots in the field of high performance computing.In recent years, many experts and researchers at home and abroad have carried out active research and exploration on multi-core processor process scheduling, aiming at improving the process scheduling efficiency through rapid process allocation and adaptive adjustment.However, the process scheduling algorithm of multi-core processors is still not ideal, and the process allocation efficiency is not high, the complexity of dynamic adjustment algorithm is high, and the load imbalance phenomenon is serious, which seriously affects the process scheduling performance of multi-core processors.Therefore, in order to find a more efficient process scheduling algorithm, this paper will focus on the multi-core processor process scheduling algorithm.Based on the analysis and study of the existing multi-core processor process scheduling algorithms, this paper proposes a comprehensive and efficient process scheduling algorithm, Adaptive Fast allocation scheduling algorithm AFASAS.AFAS scheduling algorithm, which is compared with the existing algorithms, aiming at the shortcomings of the existing efficient process scheduling algorithms.The technical advantage of this algorithm is to reduce the idle time of the processor and ensure the order of the process execution by combining the interval insertion technique with the key node algorithm.In order to improve the process allocation efficiency, the process allocation efficiency is improved by using the loop partition algorithm, and the heuristic retrieval is carried out according to the constraints of the process, and the optimization of the process allocation scheme is ensured by using the backtracking mechanism.The processor load limit value is used to adjust the load balance of the process, and the system resource overhead caused by the process migration is reduced.At the same time, for multi-process parallel allocation of multi-core processors, a specific allocation scheme is designed, which is conducive to the better application of the process allocation algorithm to practice, and to improve the performance of multi-core processor process allocation.In order to verify the feasibility and efficiency of the comprehensive and efficient process scheduling algorithm, this paper designs a reasonable performance test scheme, and analyzes its performance and results.Experimental results show that the new process allocation algorithm effectively overcomes the shortcomings of the traditional algorithms, such as low process allocation efficiency and low load balance, and effectively improves the process allocation performance of multi-core processors.It has good application prospect and theoretical research value.
【學位授予單位】：哈爾濱工程大學
【學位級別】：碩士
【學位授予年份】：2012
【分類號】：TP332

【參考文獻】

相關(guān)期刊論文 前10條

1 楊際祥;譚國真;王榮生;;多核軟件的幾個關(guān)鍵問題及其研究進展[J];電子學報;2010年09期

2 王淼;王志英;鄔貴明;;一個面向異構(gòu)多核處理器Cell的資源分配模型[J];高技術(shù)通訊;2010年12期

3 扈嘯;陳書明;;面向多核片上Trace數(shù)據(jù)流合成的隊列調(diào)度算法設(shè)計及實現(xiàn)[J];計算機研究與發(fā)展;2008年03期

4 李仁發(fā);劉彥;徐成;;多處理器片上系統(tǒng)任務(wù)調(diào)度研究進展評述[J];計算機研究與發(fā)展;2008年09期

5 章軍,馮秀山,韓冀中,韓承德;總線互連機群系統(tǒng)上的靜態(tài)任務(wù)調(diào)度[J];計算機研究與發(fā)展;1999年07期

6 孫利榮,蔣澤軍,王麗芳;片上網(wǎng)絡(luò)[J];計算機工程;2005年20期

7 何琨;趙勇;黃文奇;;基于任務(wù)復制的分簇與調(diào)度算法[J];計算機學報;2008年05期

8 劉必慰;陳書明;汪東;;先進微處理器體系結(jié)構(gòu)及其發(fā)展趨勢[J];計算機應(yīng)用研究;2007年03期

9 謝向輝;胡蘇太;李宏亮;;多核處理器及其對系統(tǒng)結(jié)構(gòu)設(shè)計的影響[J];計算機科學與探索;2008年06期

10 周佳祥,鄭緯民;基于DAG圖解-重構(gòu)的機群系統(tǒng)靜態(tài)調(diào)度算法[J];軟件學報;2000年08期

相關(guān)碩士學位論文 前5條

1 劉侃侃;異構(gòu)多核環(huán)境下的任務(wù)調(diào)度方法研究[D];杭州電子科技大學;2010年

2 林明煌;多核處理器原型驗證平臺的研究與實現(xiàn)[D];北京交通大學;2008年

3 覃中;基于多核系統(tǒng)的線程調(diào)度[D];電子科技大學;2009年

4 周升;多種片上網(wǎng)絡(luò)拓撲結(jié)構(gòu)性能評估[D];浙江大學;2010年

5 鄭凱;對數(shù)據(jù)在異構(gòu)多核處理器模擬器中進行任務(wù)劃分的研究[D];上海交通大學;2008年

，

本文編號：1745976