慶應義塾大学
2016年度春学期

コンピューター・アーキテクチャ
Computer Architecture

担当: Rodney Van Meter
E-mail: rdv@sfc.keio.ac.jp

概要

The lectures for this class will be in English, but Japanese-speaking students are encouraged to join. Adequate materials to learn the topic and complete the assignments will be available in both languages.

ムーアの法則の後期に入ったということは、先端のマルチコア・マイクロプロセッサは新しい時代を迎えたということだ。これからのコンピューターデザインはその事実をふまえなければなりません。並列データ処理、データの転送、熱等は非常に重要なファクターです。現在のコンピュータ環境を支えるマイクロプロセッサの技術を中心に扱い、動作原理や高速化の原理について考えます。次のような課題を扱います。

計算機の構成と動作原理
高級言語と機械語
コンパイラと最適化技術
オペレーティングシステムとプロセッサの関係
プロセッサの構成と設計
インプット・アウトプットシステム、データストレージシステム
マルチプロセッサネットワーク

以下の事項を注意してください。
内容を理解し、課題を行う上で基本的なC言語の知識が必要となる場合があります。しかし、大きなプログラムや複雑なプログラムを作成する能力は必要ありません。C言語の文法 (繰り返し、条件判断、関数呼出し、記憶クラスなど) の知識だけで十分です。したがって、Java 言語や Pascal 言語などの他の手続き型言語しか知らなくても、数時間の自習で理解できます。

We have entered the late Moore's Law period of computer architecture, as shown by the latest multicore microprocessors. From this point forward, computer design must concentrate on parallel processing, data transfer, and heat management as especially important factors. Studying modern microprocessor design, we will see where performance is created in computer systems, including:

Structure and development of processors
High-level language and machine language
Compiler optimization
Relationship of architecture and operating systems
Input/output systems, data storage systems
Multiprocessor networks

Note:

Students will be expected to understand at least one high-level language as part of the course. Lectures will be in English, but Japanese speaking students are encouraged to join the class. Sufficient materials will be available in both English and Japanese to learn the material and complete the assignments.

教科書 Textbook

The textbooks are available in the Seikyo:

コンピュータの構成と設計~ハードウエアとソフトウエアのインタフェース第3版 (上)
デイビッド・A. パターソン (著), ジョン・L. ヘネシー (著), David A. Patterson (原著), John L. Hennessy (原著), 成田光彰 (翻訳)
日経BP社
ISBN-10: 482228266X
ISBN-13: 978-4822282660 and the second volume: コンピュータの構成と設計~ハードウエアとソフトウエアのインタフェース第 3版 (下)
デイビッド・A. パターソン (著), ジョン・L. ヘネシー (著), David A. Patterson (原著), John L. Hennessy (原著), 成田光彰 (翻訳)
日経BP社
ISBN-10: 4822282678
ISBN-13: 978-4822282677

Computer Organization and Design: The Hardware-Software Interface, third edition
David A. Patterson and John L. Hennessy
Morgan Kauffman, 2007
ISBN-10: 0123706068
ISBN-13: 978-0123706065

A recommended, but not required, more advanced text:

コンピュータアーキテクチャ定量的アプローチ第4版
John L. Hennessy (著), ジョン・L・ヘネシー (著), デイビッド・A・パターソン (著), David A. Patterson (著), 中條拓伯 (監修, 翻訳), 吉瀬謙二 (翻訳), 佐藤寿倫 (翻訳), 天野英晴 (翻訳)
翔泳社
ISBN-10: 4798114405
ISBN-13: 978-4798114408 Computer Architecture: A Quantitative Approach, fourth edition
John L. Hennessy and David A. Patterson
Morgan Kauffman, 2007
ISBN: 978-0-12-370490-0

教材

Lecture by Lecture

第1回速い　
Lecture 1: Fast

配布資料 Lecture Notes
Lampson, Hints for Computer System Design

Lecture 1: Fast
  Concepts: What's in a computer?
  	    What's our goal in designing a computer?
	    Reading a performance graph
  In-class exercise: show around Intel CPUs, computer parts
  Homework (programming/hands-on computing): install R on your laptop
  Homework (analysis problem set): (none)

第2回 Five Principles of Computer Architecture
Lecture 2: Five Principles of Computer Architecture

配布資料 Lecture Notes

Lecture 2: Five principles of computer architecture
  Concepts: Take Advantage of Parallelism
  	    Principle of Locality
	    Focus on the Common Case
    	    Amdahl's Law
    	    The Processor Performance Equation
  In-class exercise: recreate performance graph
  Homework (programming): extending performance graphs
  Homework (analysis problem set): (none)

第3回 and 第4回：並列
Lecture 3 and Lecture 4: Parallelism, and This is content for both lectures 3 and 4.

配布資料 Lecture Notes

Lecture 3: Faster: Parallelism
  Concepts: Impact of Amdahl's Law
  	    Reduction
	    Synchronization
  In-class exercise: Parallel humans
  Homework (programming): Berkeley Parallel Boot Camp exercise
  	   		  (n.b.: single hardest exercise this semester!)
  Homework (analysis problem set): (none)

第3回 and 第4回：並列
Lecture 3 and Lecture 4: Parallelism, and This is content for both lectures 3 and 4.

配布資料 Lecture Notes

Lecture 4: Processors: Basics of Instruction Sets
  Concepts: Main Blocks of a CPU
  	    Memory: Registers, Stacks and Heaps
	    Types of Instructions
  In-class exercise: (none)
  Homework (programming): install the MIPS simulator (must be done on time!)
  Homework (analysis problem set): matrix multiplication

第5回命令の基本 Lecture 5: Processors: Basics of Instruction Sets

配布資料 Lecture Notes

Lecture 5: Processors: Instruction Sets and the Data Path
  Concepts: Details of instructions
  In-class exercise: using the MIPS simulator
  Homework (programming): matrix multiplication in C
  Homework (analysis problem set): (none)

第6回　パイプライン Lecture 6: Pipelining

配布資料 Lecture Notes

Lecture 6: Processors: Basics of Pipelining
  Concepts: Pipelines in the real world
  	    A five-stage pipeline
  In-class exercise: Human Pipeline!
  Homework (programming): matrix multiplication in MIPS assembly (due
  Dec. 20)
  Homework (analysis problem set): a real-world pipeline

第7回 Memory: Caching and Memory Hierarchy Lecture 7: Memory: Caching and Memory Hierarchy

配布資料 Lecture Notes

Lecture 7: Memory: Caching and Memory Hierarchy
  Concepts: Four key questions in cache design
  In-class exercise: (none)
  Homework (programming): (none)
  Homework (analysis problem set): cache behavior

第8回仮想メモリ Lecture 8: Virtual Memory

配布資料 Lecture Notes

Lecture 8: Memory: Virtual Memory
  Concepts: Address translation
  In-class exercise: (none)
  Homework (programming): (TBD)
  Homework (analysis problem set): (TBD)

第9回共有メモリ並列システム Lecture 9: Shared-Memory Multiprocessors

配布資料 Lecture Notes

Lecture 9: Systems: Shared-Memory Multiprocessors 
  Concepts: Classes of parallel architectures
  In-class exercise: review of Boot Camp exercise
  Homework (programming): (none)
  Homework (analysis problem set): (none)

第10回分散メモリ並列システム Lecture 10: Distributed-Memory Multiprocessors

配布資料 Lecture Notes

Lecture 10: Systems: Distributed-Memory Multiprocessors
  and Interconnection Networks
  Concepts:
  In-class exercise: (none)
  Homework (programming): (TBD)
  Homework (analysis problem set): (TBD)

第11回算数 Lecture 11: Arithmetic

配布資料 Lecture Notes

Lecture 11: Arithmetic
  Concepts: integer arithmetic
  	    floating point arithmetic
  In-class exercise: (none)
  Homework (programming): precision and performance of floating point
  	   		  arithmetic
  Homework (analysis problem set): (TBD)

第12回入出力
Lecture 12: Basics of I/O and Storage Systems

Lecture 12: Basics of I/O and Storage Systems
  Concepts: disk drive internals
  	    impact of bus transfers on performance
  In-class exercise: (none)
  Homework (programming): (TBD)
  Homework (analysis problem set): (TBD)

第13回 RAID Lecture 13: RAID

配布資料 Lecture Notes

Lecture 13: RAID: Parallelism and Protection in Storage Systems 
  Concepts: RAID levels
  In-class exercise: (none)
  Homework (programming): (TBD)
  Homework (analysis problem set): (TBD)

第14回ファステスト
Lecture 14: Fastest: Ultimate limits to computation

配布資料 Lecture Notes

Lecture 14: Fastest: Ultimate Limits to Computation
  Concepts: TBD
  In-class exercise: (TBD)
  Homework (programming): (TBD)
  Homework (analysis problem set): (TBD)

Requirements

The course consists of fourteen ninety-minute classes. Students are expected to read a large amount of material from the textbook and accompanying CD, contribute to classroom discussions, complete the homeworks, and pass a final exam.

There are several types of exercises that you must do during the course of this semester. You grade will be composed of the following, with a maximum total of 250 points:

In-class exercises: some hands-on things in class, some of which involve programming (combined with attendance and participation, 50 points)
Programming/hands-on (100 points):
1. Install R (trivial)
2. Measure parallel performance (easy)
3. Berkeley parallel boot camp (hard, double credit)
4. Install MIPS simulator (trivial)
5. Matrix multiplication in C (easy)
6. Matrix multiplication in assembler (hard, double credit)
7. Integer performance, floating point precision and performance (easy)
8. TBD
Analysis problem sets (50 points):
1. TBD
2. Matrix multiplication/counting computational operations
3. Estimating CPU and memory performance
4. Optimizing in the context of design constraints (hard, double credit) (last exercise, due end of the semester)
Final exam (50 points)

Homework is due at the start of class each week. Homework will be submitted via the SFS system.

その他