- Date: September 28th, 2016
- Time: 10:00AM-12:00PM
- Location: 116 Pearson Hall, Faculty Commons
Using parallel programming models for migrating legacy code to today’s platforms
Presenters: Prof. Sunita Chandrasekaran, Sergio Pino (third year graduate student) and Collion Clark (undergraduate student).
Assistant Professor Sunita Chandrasekaran, Computer and Information Sciences, and her research team led the September 2016 session of the High Performance Computing (HPC) Symposium Series on September 28.
Sunita began her presentation with a discussion about the benefits of UD becoming an NVIDIA GPU Education Center, which Sunita helped UD attain with her co-organization of the 2016 GPU Hackathon. Chandrasekaran went on to discuss how she uses her nodes on Farber to understand state-of-the-art hardware, to explore compilation techniques, to run applications faster, and to learn to use new evaluation metrics. Chandrasekaran’s use of Farber demonstrates the benefits of utilizing GPUs and parallel programming to speed up an application’s run time.
The presentation ended with a discussion about the merits of OpenACC and OpenMP and the plans for the forthcoming cluster. Researchers are encouraged to use this form to indicate their needs for the next cluster. During the open-forum, the group discussed how using UD’s community clusters can be a stepping stone to earning larger grants and access to larger clusters.
For those interested in learning more about OpenACC, a free online course hosted by NVIDIA will begin October 26th. The introduction to OpenACC will help you start accelerating your codes with GPUs through lectures, hands-on labs, and time with experts.
