In this paper we study the effects of using multiple-subscripted arrays in benchmark programs. We identify and exemplify typical problems caused by multiple-subscripted arrays and show why their usage in benchmarking should be strictly controlled and frequently restricted. Multiple-subscripted arrays can be considered harmful in the case of general purpose processor-bound benchmarks. On the other hand, the multiple-subscripted arrays are shown to be suitable for measuring the optimizing features of compilers, especially for RISC machines.
/lp/association-for-computing-machinery/the-use-of-multiple-subscripted-arrays-in-benchmark-programs-2pVwLnN0S6