Locality is an algorithm characteristic describing a usage level of fast access memory. For example, in case of distributed memory computers we focus on memory of each computational node. To achieve the high performance of algorithm implementation one should choose the best possible locality option. Studying the parallel algorithm locality is to estimate the number and volume of data communications. In this work, we formulate and prove the statements for computers with distributed memory that allow us to estimate the asymptotic volume of data communication operations. These estimation results are useful while comparing alternative versions of parallel algorithms during data communication cost analysis.
A floating-point validation suite for high-performance shared and distributed memory computing systems