Understanding parallel computing techniques is becoming increasing important….
To support teaching the topics, tools, and techniques effectively, VU has invested in a number of specialized assets for student projects and classroom that were inspired by the work of the CSinParallel Community.
These included 3 clusters, based on:
- the NVIDIA Jetson TK-1 (6 Nodes),
- the LittleFe v3d (with 6 Nodes), and
- the Parallella single-board system (5 nodes).
Since the initial deploy of those three systems, four additional clusters have been added:
- 2 based on the Raspberry Pi 3B Plus (5 nodes) and
- 2 based on the ODROID XU4
This second generation is also designed to allow an easy combination of the twin 5 node clusters into a single 10 node cluster.
All of the systems were built by students, and have been used to support classroom and independent study projects on parallel, distributed, and high performance computing. The Raspberry Pi systems have also been the basis of a partnership with the Department of Meteorology and Geography.
These assets are complemented by the more powerful compute systems, including “pascal.”
The original front end was “acorn” (Front End), which has been replaced in that role by “tweed” and “plaid”.
Many thanks are due to Justin Szaday, Paul Nord, and Terry Wade who helped build out these systems.