Chapter 1 Fundamentals of computer system¶

冯诺依曼结构（Von Neumann Structure）

Flynn 分类法

注：IS=指令流，DS=数据流，CS=控制流，CU=控制单元，PU=处理单元

Measuring Data Size

单位	换算关系
bit	1 位（二进制位）
nibble	4 bits
byte	8 bits
word	嵌入式/移动处理器：4 bytes (32 bits) 桌面/服务器：8 bytes (64 bits)
kibibyte (KiB) [kilobyte (KB)]	\(2^{10}\) (1,024) bytes
mebibyte (MiB) [megabyte (MB)]	\(2^{20}\) (1,048,576) bytes
gibibyte (GiB) [gigabyte (GB)]	\(2^{30}\) (1,073,741,824) bytes
tebibyte (TiB) [terabyte (TB)]	\(2^{40}\) (1,099,511,627,776) bytes
pebibyte (PiB) [petabyte (PB)]	\(2^{50}\) (1,125,899,906,842,624) bytes

CPU 性能回顾

\[ Performance = \frac{1}{Execution\ time} \]

\[ \frac{Performance_X}{Performance_Y} = \frac{Execution\ time_Y}{Execution\ time_X} = n \]

\[ CPU\ Time = CPU\ Clock\ Cycles × Clock\ Cycle\ Time = \frac{CPU\ Clock\ Cycles}{Clock\ Rate} \]

\[ CPI = \frac{CPU\ Clock\ Cycles}{Instruction\ Count} \]

\[ CPU\ Time = IC × CPI × Clock\ Period = \frac{IC × CPI}{Clock\ Rate} \]

\[ CPI = \sum_{i=1}^n (CPI_i × \frac{IC_i}{IC}) \]

Amdahl's Law

\[ T_{new} = T_{old} × \left[(1-F_{enhanced}) + \frac{F_{enhanced}}{S_{enhanced}}\right] \]

\[ Speedup_{overall} = \frac{T_{old}}{T_{new}} = \frac{1}{(1-F_{enhanced}) + \frac{F_{enhanced}}{S_{enhanced}}}< \frac{1}{1-F_{enhanced}} \]