以品質導向之快速系統回復的線上頁面選擇方法

Abstract

隨著可攜式電子產品的流行，系統休眠與恢復發揮不可或缺的功用，讓使用者可以在不使用時關閉電源並恢復原來使用狀態。然而掛載休眠映像為大部份系統的效能瓶頸。在本論文中，我們提出一個新的記憶體管理系統架構，將記憶體頁面延後至掛載休眠映像後載入，並保證恢復後的系統品質與效能。我們提出兩個近似演算法去減少休眠映像大小並證明在最糟情況下演算法所產生之記憶體頁面排程，兩者皆有大於等於最佳解之一半。從實驗數據中，在考慮/不考慮頁面死線限制之下，相對於Linux作業系統舊有機制各別有22.08%與9.34% 改進。

With increasing popularity of mobile devices recently, there has been growing interest in providing a better user experience. In a hibernation system, good user experiences are fast resuming to the first available use and responsiveness for user applications. In this paper, we proposes a memory management system to select memory pages into scheduled page sets to reduce the image loading time(which is the bottleneck of the resuming) and guarantee the resumed system quality in the mean time. Two bounded approximation algorithms are proposed to produce resulting schedules with and without time constraint respectively, and average 22.08 %, 9.34 % improvement over a Linux default setting.