Pregled bibliografske jedinice broj: 1238358
Evaluating Performance of SBC Clusters for HPC Workloads
Evaluating Performance of SBC Clusters for HPC Workloads // Proceedings of International Conference on Smart Systems and Technologies (SST 2022)
Osijek: Institute of Electrical and Electronics Engineers (IEEE), 2022. str. 173-178 doi:10.1109/sst55530.2022.9954756 (predavanje, međunarodna recenzija, cjeloviti rad (in extenso), znanstveni)
CROSBI ID: 1238358 Za ispravke kontaktirajte CROSBI podršku putem web obrasca
Naslov
Evaluating Performance of SBC Clusters for HPC Workloads
Autori
Krpic, Zdravko ; Loina, Luka ; Galba, Tomislav
Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni
Izvornik
Proceedings of International Conference on Smart Systems and Technologies (SST 2022)
/ - Osijek : Institute of Electrical and Electronics Engineers (IEEE), 2022, 173-178
ISBN
978-1-6654-8215-8
Skup
International Conference on Smart Systems and Technologies 2022 (SST 2022)
Mjesto i datum
Osijek, Hrvatska, 19.10.2022. - 21.10.2022
Vrsta sudjelovanja
Predavanje
Vrsta recenzije
Međunarodna recenzija
Ključne riječi
Single Board Computer ; SBC cluster ; High Performance Linpack ; performance evaluation
Sažetak
Clusters of modern Single Board Computers (SBCs) pack significant computing power thanks to the new generation of low-power CPUs. So far, such clusters have mainly been used for educational and hobbyist projects but since their computing power grew, they are increasingly recognized as a disruptive technology, especially in fog and edge computing. This is also due to their low cost and low power consumption, as well as their expendability. In this paper we will investigate how do such clusters cope with compute-bound applications. To do so we will use High Performance Linpack (HPL) benchmark, a well-established tool for evaluating performance of High Performance Computing systems, to compare processing power of two 4-node clusters using different generations of SBCs: Odroid U3 and Odroid MC1. The results show that clusters of SBCs are indeed small-scale HPCs capable of running moderate computing workloads that could be generated at the computing edge.
Izvorni jezik
Engleski
Znanstvena područja
Računarstvo
POVEZANOST RADA
Ustanove:
Fakultet elektrotehnike, računarstva i informacijskih tehnologija Osijek