課程資訊
課程名稱
高等計算機結構
Advanced Computer Architecture 
開課學期
106-2 
授課對象
電機資訊學院  資訊工程學研究所  
授課教師
楊佳玲 
課號
CSIE5059 
課程識別碼
922EU1470 
班次
 
學分
3.0 
全/半年
半年 
必/選修
選修 
上課時間
星期二7,8,9(14:20~17:20) 
上課地點
資310 
備註
本課程以英語授課。
總人數上限:30人 
Ceiba 課程網頁
http://ceiba.ntu.edu.tw/1062CSIE5059_ACA 
課程簡介影片
 
核心能力關聯
核心能力與課程規劃關聯圖
課程大綱
為確保您我的權利,請尊重智慧財產權及不得非法影印
課程概述

Computer architecture has evolved — from a world of mainframes, minicomputers, and microprocessors, to a world dominated by microprocessors, and now into a world where microprocessors themselves are encompassing all the complexity of mainframe computers. 

課程目標
This course focuses on advanced computer architecture design such as deep pipelining, techniques to exploit instruction level parallelism and thread level parallelism, and memory hierarchy management.
Students will acquire the skills of evaluating the performance of alternative design choices in system design. 
課程要求
待補 
預期每週課後學習時數
 
Office Hours
 
指定閱讀
待補 
參考書目
Textbook & Reference Books:
(1) Computer Architecture: A Quantitative Approach. 5th Edition, John L.
Hennessy and David A. Patterson, Morgan, 20011
(2) Selected papers 
評量方式
(僅供參考)
   
課程進度
週次
日期
單元主題
第1週
2/27  Course Introduction 
第2週
3/06  [Lecture]
Basics of Computer Architecture Design  
第3週
3/13  [Lecture]
Dynamic Instrcution Scheduling/ Branch Predictor (I) 
第4週
3/20  [Lecture]
Dynamic Instrcution Scheduling/ Branch Predictor (II) 
第5週
3/27  [Lecture]
Memory Hierarchy: Cache 
第6週
4/03  Spring Break 
第7週
4/10  [Lecture]
DRAM architecture/3D memory/NVM/SSD 
第8週
4/17  Midterm Exam 
第9週
4/24  [Term project announcement]
[Paper presentation I: Memory Hierarchy]
1.林蔚廷: Jenga:Software-Defined Cache Hierarchies(ISCA 2017)
2.吳易忠: Improving Cache Performance by Exploiting Read-Write Disparity(HPCA 2014)
3.魏禛: CREAM:A Concurrent-Refresh-Aware DRAM Memory System(HPCA 2014) 
第10週
5/01  [Lecture]
Data Parallelism/GPU/TPU
 
第11週
5/08  [Paper presentation II: NVM/SSD]
1.楊子賢: Beyond block I/O: Rethinking traditional storage primitive(HPCA 2016)
2.狄丹: Morpheus:Creating Application Objects Efficiently for Heterogeneous Computing(ISCA 2016)
3.林聰: Endurable Transient Inconsistency in Byte-Addressable Persistent B+-Tree(FAST 2018)
 
第12週
5/15  [Paper Presentation III: GPU]
1.王本奕: Dynamic Thread Block Launch: A Lightweight Execution Mechanism to Support Irregular Applications on GPUs(ISCA 2015)
2.吳哲宇: Divergence-Aware Warp Scheduling(MICRO 2013)
3.邱淳浩: Access Pattern-Aware Cache Management for Improving Data Utilization in GPU(ISCA 2017) 
第13週
5/22  Discussion on Project  
第14週
5/29  [Lecture]
Low Power 
第15週
6/05  [Paper Presentation IV: Machine learning]
1.徐彥旻: Maximizing CNN Accelerator Efficiency Through Resource Partitioning(ISCA 2017)
2.謝耀慶: In-situ AI: Towards Autonomous and Incremental Deep Learning for IoT Systems(HPCA 2018)
3.顏睿楠: vDNN:Virtualized Deep Neural Networks for Scalable, Memory-Efficient Neural Network Design(MICRO 2016) 
第16週
6/12  [Paper Presentation V: low power]
1.黃世瑋: Dynamically Exploiting Narrow Width Operands to Improve Processor Power and Performance(HPCA 1999)
2.郭桂呈: CHARSTAR:Clock Hierarchy Aware Resource Scaling in Tiled Architectures(ISCA 2017) 
第17週
6/19  Project Presentation 
第18週
6/26  Final Exam