Abstract :
The three most often used benchmarks are characterized in detail and users are warned about a number of pitfalls. Two of them, Whetstone and Drystone, are synthetic benchmarks: they were written solely for benchmarking purposes and perform no useful computation. Linpack was distilled out of a real, purposeful program that is now used as a benchmark. Some other benchmarks, namely Livermore Fortran Kernels, Stanford Small Programs Benchmark Set, EDN benchmarks, Sieve of Eratosthenes, Rhealstone and SPEC benchmarks, are briefly considered. Non-CPU influences in benchmark performance are discussed.<>
Keywords :
performance evaluation; program testing; software selection; Drystone; EDN benchmarks; Linpack; Livermore Fortran Kernels; Rhealstone; SPEC benchmarks; Sieve of Eratosthenes; Stanford Small Programs Benchmark Set; Whetstone; benchmark performance; common benchmarks; synthetic benchmarks; Application software; Computer aided instruction; Computer architecture; High level languages; Information systems; Manufacturing; Measurement; Performance evaluation; Reduced instruction set computing; Voice mail;