Performance Metrics and TargetsLatency, Throughput, Energy, and PowerPeak Performance as the Ultimate LimitScalability and Maximum Parallel SpeedupBottlenecks and a Bit of Queuing TheoryRoofline ModelPerformance Features of Computer ArchitecturesIncreasing Single-Threaded Performance: Where You Can and Cannot HelpProcess More Data with SIMD ParallelismDistributed and Shared Memory SystemsHPC Hardware Architecture OverviewA Multicore Workstation or a Server Compute NodeCoprocessor for Highly Parallel ApplicationsGroup of Similar Nodes Form an HPC ClusterOther Important Components of HPC SystemsSummaryReferences