Research Support
Mills end of life
On January 27, the HPC team announced the plans for the final year of the Mills high performance computing (HPC) cluster. The Mills cluster was designed to have a 5-year life span; that span concludes at the end of the 2016 calendar year. A statement was released that provides an overview of the HPC team's strategy for moving forward. The formal policy document describes the strategy for the retirement of Mills. Questions about Mills and its final year can be directed to the HPC team using the HPC help request form.
HPC Symposium series
The IT Research Computing group hosted the February 2016 HPC Symposium to address the questions and concerns of UD's HPC community. Over 24 attendees participated in this open forum which included a discussion of the Mills End of Life policy.
The first hour of the symposium included a formal presentation about the implementation of the spillover queue on Mills, "Mills: Job Scheduling Change." The presentation was followed by questions from the audience regarding how the spillover queue will affect their jobs and how they will need to change their job submissions. Additional questions were raised about the upcoming cluster, leading to a discussion about researcher needs and the design of the next cluster.
The remaining time was spent in breakout groups led by HPC
Team members discussing "Benchmarking jobs: Computational models
for running Matlab on a shared cluster," "Getting started with
Farber," and "Troubleshooting on Mills and Farber."
The next HPC Symposium will be held on June 8 featuring the Neunuebel Neuroscience Lab. (Register)
2016 UD GPU Hackathon
Professor Sunita Chandrasekaran, Computer and Information Sciences, hosted a GPU Programming Hackathon at UD, May 2-6, to help developers with scalable applications to port to, or optimize, their applications on a GPU accelerator. National labs, universities, and vendors--including OpenACC, Cray, NVIDIA, Oak Ridge National Laboratory, and PGI--were present to mentor development groups. Members of the IT HPC team supported Chandrasekaran in coordinating the Hackathon and provided technical support to the teams during the event.
