Applications for 68 Node Raspberry Pi 3 Education Cluster Kelley A. Deuso, Gregory Simpson, Alok Hota, Mohammad Raji, Aaron R.Young, J. Parker Mitchell, Jordan Sangid, and Mark Dean The University of Tennessee, Knoxville TN 37996, USA markdean@utk.edu Abstract. In order to provide students access to a low-cost high perfor- mance computing cluster, the Big Orange Bramble (BOB) project was created. The project involved the design and implementation of a 68 node Raspberry Pi 3 cluster, for which Pi calculation, Monte Carlo sim- ulation, fire dynamics simulation, and Spark applications were developed and installed to demonstrate the educational advantages associated with distributed and parallel computing education. Keywords: cluster · HPC · ARM · education · Spark 1 Introduction There is an increasing need for computational sciences education to include a focus on distributed and parallel computing, particularly with the rise of cloud computing and big data processing. For students, the prevailing limitations of hands-on access to high performance computing clusters are their associated cost, housing, and maintenance. In order to address this need, the Big Orange Bramble (BOB) project was conceived to design a low-cost, easy-to-setup Be- owulf cluster that students could develop, maintain, and utilize to further their parallel computing education. The project involved the design and construction of a cluster composed of 68 quad-core ARMv8 64-bit Raspberry Pi 3s. Aside from building the cluster, the primary goals of the project were to establish the operating environment, communication structure, application frameworks, appli- cation development tools, and libraries to support a high performance cluster. Additionally, open documentation was created to provide information to anyone wishing to copy or expand upon BOB’s efforts. This paper seeks to expound the development, analysis, and educational impact of the initial applications created on BOB. 1.1 Hardware Overview In order to understand the types of applications which could be developed, it is important to note the major hardware components and their relevant limitations.