Performance Analysis of a Low Cost Cluster with Parallel Applications and ARM Processors
Francisco Aldiel Lima, Edward David Moreno, Wanderson Roger Azevedo Días · IEEE Latin America Transactions · 2016
This paper presents the results obtained during the simulations which measured the performance of a low-cost cluster with ARM processors. For the simulations, we consider the impact caused by parallel applications responsible for big chunks of calculations (HPL benchmark, matrix multiplication and scalar product). Thus, after the simulations, we could observe that the cluster performance increased considerably when compared to sequential solutions, and this good speedup is better when the measured amount of processed data also increased. For matrix multiplication, the gain came to 66% when compared to the sequential solution.