计算机组成与结构：chapter18 Parallel Processing.pptVIP

Computer Organization ArchitectureChapter 18Parallel Processing 18.1 Multiple Processor OrganizationClassification of parallel processor system based on FlynnSingle instruction, single data stream - SISDSingle instruction, multiple data stream - SIMDMultiple instruction, single data stream - MISDMultiple instruction, multiple data stream - MIMD SISDSingle processorSingle instruction streamData stored in single memoryUni-processorPipeline SIMDSingle machine instruction controls simultaneous execution of a number of processing elementsLockstep basedEach processing element has associated data memoryEach instruction executed on different set of data by different processorsVector and array processors MISDSequence of dataTransmitted to set of processorsEach processor executes different instruction sequenceNever been implemented MIMDSet of processorsSimultaneously execute different instruction sequencesDifferent sets of dataSMPs, clusters and NUMA systems A Taxonomy of Parallel Processor Architectures MIMD - OverviewGeneral purpose processorsEach can process all instructions necessaryFurther classified by method of processor communicationShared memorySMP– symmetric multiprocessorNUMA– non-uniform memory accessDistributed memoryclusters Tightly Coupled - SMPProcessors share memoryCommunicate via the shared memorySymmetric Multiprocessor (SMP)Share single memory or pool of memoryBy means of shared bus or other interconnection mechanism to access memoryMemory access time to given area of memory is approximately the same for each processor Tightly Coupled - NUMANon-uniform memory accessAccess times to different regions of memory may differ for a NUMA Loosely Coupled - ClustersCollection of independent uniprocessors or SMPsInterconnected to form a clusterCommunication via fixed path or network connections Parallel Organizations - SISDCU– control unitPU– processing unitMU– memory unitIS– instruction streamDS– data stream Parallel Organizations - SIMDPE—processing el