计算机体系结构chapter06.pptx

计算机体系结构周学海mailto:xhzhou@xhzhou@0551国科学技术大学中国科学技术大学第6章 Data-Level Parallelism in Vector, SIMD, and GPU ArchitecturesSIMD的变体向量体系结构GPU计算机体系结构review: MultithreadingSimultaneousMultithreadingMultiprocessingSuperscalarFine-GrainedCoarse-GrainedTime (processor cycle)Thread 1Thread 3Thread 5Thread 2Thread 4Idle slot传统指令级并行技术的问题挖掘ILP的传统方法的主要缺陷: 提高流水线的时钟频率: 提高时钟频率，有时导致CPI随着增加 (branches, other hazards) 指令预取和译码: 有时在每个时钟周期很难预取和译码多条指令提高Cache命中率 : 在有些计算量较大的应用中（科学计算）需要大量的数据，其局部性较差，有些程序处理的是连续的媒体流(multimedia),其局部性也较差。IntroductionSIMD 结构可有效地挖掘数据级并行:基于矩阵运算的科学计算图像和声音处理SIMD比MIMD更节能针对每组数据操作仅需要取指一次SIMD对PMD( personal mobile devices)更具吸引力SIMD 允许程序员继续以串行模式思维SIMD Parallelism向量体系结构SIMD 扩展Graphics Processor Units (GPUs)For x86 processors:每年增加2cores/chipSIMD 宽度每4年翻一番SIMD潜在加速比是MIMD的2倍SupercomputersDefinition of a supercomputer:对于给定任务而言世界上最快的机器将计算密集型转化为I/O密集型的设备任何造价超过3千万美元的机器由Seymour Cray设计的机器CDC6600 (Cray, 1964) 被认为是第一台超级计算机`CDC 6600 Seymour Cray, 1963A fast pipelined machine with 60-bit words128 Kword main memory capacity, 32 banksTen functional units (parallel, unpipelined)Floating Point: adder, 2 multipliers, dividerInteger: adder, 2 incrementers, ...Hardwired control (no microcoding)Scoreboard for dynamic scheduling of instructions Ten Peripheral Processors for Input/Outputa fast multi-threaded 12-bit integer ALUVery fast clock, 10 MHz (FP add in 4 clocks)400,000 transistors, 750 sq. ft., 5 tons, 150 kW, novel freon-based technology for coolingFastest machine in world for 5 years (until 7600)over 100 sold ($7-10M each)3/10/2009IBM Memo on CDC6600Thomas Watson Jr., IBM CEO, August 1963: “Last week, Control Data ... announced the 6600 system. I understand that in the laboratory developing the system there are only 34 people including the janitor. Of these, 14 are engineers and 4 are programmers... Contrasting this modest effort with our vast development activities, I fail to understand why we have lost our industry leadership position by letting someone else offer the worlds