Computer Composition and Design 01 - Computer Overview and Technology

Table of contents

Series Article Directory

1. Content of this course

2. Eight great ideas of computer system architecture

1. Moore's Law

2. Abstraction and simplification

3. Accelerate high-probability events

4. Abstraction – simplifying the design

5.API vs. ABI

6. Manufacturing of ICs

7. Computer efficiency

8. Number of instructions and CPI

8.1 CPI

8.2 Complex CPI

8.3 Performance Metrics

9. Power Consumption Trend

10. Some pitfalls

10.1 Improvement (needs analysis) 

10.2 Power Consumption and Utilization

10.3 MIPS 

Summarize


Series Article Directory

This series of blogs focuses on the core content of the computer system (3) course of Shenzhen University, and the bibliography "Computer Composition and Design" (if you have any questions, please discuss and point out in the comment area, or contact me directly by private message).

Chapter 1 Computer Composition and Design 01 - Computer Overview and Technology


synopsis

This blog mainly introduces the core content of the computer composition and design book, as well as the relevant knowledge of computer overview and technology in the first chapter of the bibliography.


1. Content of this course

2. Eight great ideas of computer system architecture

1. Moore's Law

Tip: When designing, consider the situation of the integrated circuit in the future (when it goes on the market)

2. Abstraction and simplification

3. Accelerate high-probability events

Core: If the code is used iteratively, it is necessary to optimize the high probability event (important program)

4. Abstraction – simplifying the design

5.API vs. ABI

API source code compatible
ABI – binary compatible

6. Manufacturing of ICs

The purpose of multiple tests after patterning and dicing: to improve the area utilization of blank wafers

7. Computer efficiency

The efficiency of a computer is related to response time and throughput .

Response time (execution time): The total time a computer takes to complete a task

Throughput rate (bandwidth): the number of tasks completed per unit of time

  For example, number of tasks / number of transactions /... per hour

How to positively affect response time and throughput

1. Replace the processor with a higher speed model

2. Add multiple processors to handle independent tasks separately

Tips: We focus on response time...

Total response time includes everything (CPU time too):  processing , I/O operations , OS overhead , idle time

Clock cycle time = 1/clock frequency

    Clock cycle: The time of one clock cycle of the computer eg, 250ps = 0.25ns = 250×10–12s

    Clock frequency: number of clock cycles per second eg, 4.0GHz = 4000MHz = 4.0×109Hz 

A sample is as follows:

8. Number of instructions and CPI

8.1 CPI

CPI (Clock cycle per instruction): The average number of clock cycles required for each instruction

Tips: The number of instructions for a program depends on the program, ISA and compiler

Average clock cycles per instruction (CPI)

Determined by CPU hardware

If different instructions have different CPI , then the average CPI is affected by the entire instruction set 

    Tips: According to the CPU time ratio, the ratio of execution time (efficiency) can be derived 

8.2 Complex CPI

If an instruction sequence has n different types of instructions, each requiring a different number of clock cycles (corresponding to the definition of CPI), then 

A sample is as follows:

8.3 Performance Metrics

The performance of n programs depends on:
n Algorithm : Affects number of instructions , possible CPI
nProgramming language : Affects instruction count , CPI
Compiler : Affects the number of instructions , CPI
n instruction set architecture : affects the number of instructions , CPI, clock frequency

Examples are as follows (n-core is not n-fold higher than single-core, and there are losses):

9. Power Consumption Trend

The power consumption per transistor based on CMOS integrated circuit technology is as follows: 

 Tips: Power consumption is proportional to capacitance, voltage, and switching frequency (you can’t reduce the formula infinitely)

10. Some pitfalls

10.1 Improvement (needs analysis) 

Example : A program takes 100 seconds to run, and the multiplication operation takes 80 seconds

How much should the speed of the multiplication be improved if the program runs 5 times faster ?

10.2 Power Consumption and Utilization

Fallacy : Low Utilization Computers Use Low Power

10.3 MIPS 

MIPS is a supercomputer with millions of instructions per second. It cannot be used as a single performance indicator because it cannot explain the differences in the instruction sets between computers and the differences in the complexity of instructions.

Summarize

The above is the summary of the core content of the book and the first chapter - the core knowledge of computer overview and technology. In the first chapter, it mainly focuses on the introduction of 8 design ideas in the computer system structure , computer efficiency and power consumption .

Guess you like

Origin blog.csdn.net/weixin_51426083/article/details/126719596