透過您的圖書館登入
IP:3.147.65.65
  • 學位論文

使用基因演算法之顯示卡記憶體參數設定

Parameter Determination for Graphic Card Memory Using Genetic Algorithm

指導教授 : 王順源 曾傳蘆
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


在電腦設備中,顯示卡具有相當重要的地位,如果記憶體參數設定非最佳化,往往造成電腦顯示不穩定甚至於當機。由於一般顯示卡記憶體參數至少有十個以上且必須經由工程師以經驗來判斷,並不斷地經過嘗試錯誤後修正設定,會使顯示卡設計耗時及成本提高。 因此,本論文提出利用基因演算法來設定顯示卡記憶體參數,可有效地縮短研發時間,調整出比較穩定的參數設定值,以增加產品上市的競爭力。本論文所使用的顯示卡為NVIDIA 8600系列,有16個記憶體參數,結合基因演算法程式與生產顯示卡時所使用的測試程式,發展出自動化參數設定程式,並且在三種不同的記憶體操作頻率下,分別搜尋最佳的記憶體參數設定值群體,然後進行穩定性測試,以決定最後的參數設定值。 由實驗結果顯示,基因演算法可以縮短找尋參數設定值的時間,全自動化的設定程式可以減少工程師人力成本,不但可以提高顯示卡在工廠生產的良率,增加公司的利潤,也提高了顯示卡在超頻的狀態下,3D運行的穩定度。

並列摘要


Among computer components, the graphic card plays a quite important role. If the parameters of graphic card memory are not well-tuned and optimized, it causes the computer system unstable and even serious halt. In general, the number of parameters of graphic card memory is more than 10 and these parameters are tuned by hands-on engineers via trial and error process. The time consumption and labor cost are high for graphic card memory setting. To solve the problem, this thesis utilizes the genetic algorithm to determine the parameters of the graphic card memory. It can shorten the research and develop time effectively, and fine tune stable parameters of memory and hence increase the products competitiveness for time to market. The graphic cards used in this thesis are NVIDIA 8600 series. By integrating the genetic algorithm program and the graphic card test program, thesis proposes an automatic parameter setting system. It searches the parameters in three different operating frequencies respectively and finds out the final memory parameters by stability burn-in test. The experimental results show that the genetic algorithm can shorten the time of parameter setting. The automatic setting system effectively reduces the manpower cost of engineers, improve production yield rate in the factory, and increase the company profits. In addition, it can improve the stability of graphic card in 3D operation under the exceeded frequency mode.

參考文獻


[22] 蔡旭龍,以混合式基因演算法訓練之類神經診斷系統,碩士論文,國立台北科技大學,台北,2005。
[24] 陳肇宗,權重型基因之基因法則於特徵選取之應用,碩士論文,國立台北科技大學,台北,2006。
[26] 吳承翰,基因演算法與模擬退火法於MC-CDMA系統聯合通道估測及多用戶偵測之應用研究,碩士論文,國立台北科技大學,台北,2009。
[1] K. Iba, "Reactive power optimization by genetic algorithm," IEEE Trans. on Power Systems, vol. 9, no. 2, 1994, pp. 685-692.
[2] K. Y. Lee and F. F. Yang, "Optimal reactive power planning using evolutionary algorithms: a comparative study for evolutionary programming, evolutionary strategy, genetic algorithm, and linear programming," IEEE Trans. on Power Systems, vol. 13, no. 1, 1998, pp. 101-108.

延伸閱讀