본문으로 건너뛰기

Computer Abstractions and Technology

Moore's Law​

하나의 싱글 칩에 들어갈 수 있는 트랜지스터의 수가 매년 2배씩 늘어나는 법칙

What You Will Learn​

  • How Programs are translated into the machine language
  • The hw/sw interface
  • What determines program performance
  • How hardware designers imporve performance

Eight Great Ideas in Computer Architecture​

  1. Design for Moore's Law
  2. Use abstraction to simplify design design
  3. Make the common cast fast
  4. Performance via parallelism
  5. Performance via pipelining
  6. Performance via prediction
  7. Hierarchy of memories
  8. Dependability via redundancy (redundancy를 통한 안정성 / 신뢰성 획득)

Below Your Program​

  • Application software
    • high-level language로 작성됨
  • System software
    • Compiler: HLL code > machine code로 translate
    • Operating System
  • Hardware
    • Processor, memory, I/O controllers

Levels of Program Code​

  • High-level language
    • Level of abstraction closer to problem domain
  • Assembly language
    • Textual representation of instructions
  • Hardware representation
    • Binary digits (bits)
    • Encoded instructions and data

Under the Covers​

Inside the Processor(CPU)​

  • Datapath: performs operations on data
  • Control: sequences datapath, memory, ...
  • Cache memory
    • Small fast SRAM memory for immedicate access to data

A Safe Place for Data​

  • Volatile main memory(휘발성 메인 메모리, DRAM)
    • Loses instructions and data when power off
  • Non-volatile secondary memory memory(저장장치, storage)
    • Magnetic disk
    • Flash Memory
    • Optical

Networks​

  • Communication, resource sharing, nonlocal access
  • Local area network (LAN): Ethernet
  • Wide area network (WAN): the Internet
  • Wireless network: WIFI, Bluetooth

Technologies for Building Processors and Memory​

Performance​

Defining Performance​

  • 어떤 항공기가 가장 좋은 performance를 가지고 있나?

computer_abstractions_technology_1

  • performance를 어떻게 정의하냐에 따라 다르기 때문에, performance를 잘 정의하는 것이 중요하다.

Response Time and Throughput​

  • Response time(latency)
    • How long it takes to do a task(작업 시간)
  • Throughput
    • Total work done per unit time(단위 시간 당 작업의 양)
      • e.g., tasks/transactions/... per hour (시간당 task, transaction 수)
  • How are response time and throughput affected by
    • Replacing the processor with a faster version? : responsetime 감소, throughput 증가
    • Adding more processors? : throughput만 증가
  • We'll focus on response time for now... (지금 시점부터는 response time에 집중)

Relative Performance(상대)​

  • Define Performance = 1 / Execution Time
  • "X is n time faster than Y"
  • Performance(x) / Performance(y) = Execution time(y)/Execution time(x) = n
  • Ex. time taken to run a program
    • 10s on A, 15s on B
    • Execution Time(B) / Execution Time(A) = 15 / 10 = 1.5
    • S A is 1.5 times faster than B

Measuring Execution Time​

  • Elapsed time
    • Total response time, including all aspects(아래를 다 고려한 전체 걸리는 시간)
      • Processing, I/O, OS overhead, idle time
    • Determines system performance
  • CPU time
    • Time spent processing a given job(주어진 작업을 실제 실행하는 시간)
      • Discounts I/O time, other jobs' shares
    • Comprises user CPU time and system CPU time
    • Diffrent programs are affected differentyl by CPU and system performance

CPU Clocking​

  • Operation of digital hardware governed by a constant-rate clock

computer_abstractions_technology_2

  • 하나의 Clock period는 rising edge, falling edge 하나로 구성되어있음

  • Clock period: duration of a clock cycle

  • Clock frequency(rate): cycles perseconds

  • CC = 1 / CR

중간 정리​

  • Performance는 throughput과 response time으로 정의된다.

CPU Time​

computer_abstractions_technology_3

  • Performance improved by
  • Reducing number of clock cycles
  • Increasing clock rate
  • Hardward designer must often trade off clock rate against cycle count(clock rate는 cycle count와 비례)