May 19, 2018
The Vertically Integrated Project (VIP) program is a multi-year research as well as credit-based program for undergraduate students. Four students, Eric, Mauricio, Thomas and Edwin of CIS/CPEG recently won the VIP Best Poster Award in a inter-regional competition held at UDEL. They took a chemical shift prediction application called PPM_One, and accelerated it on state-of-the-art GPUs using the directive-based programming standard OpenACC. Results showed that the a dataset of ~2 million atoms running sequential for over 10+ hours took only 2 minutes on NVIDIA V100 GPUs.
The chemical shift of a protein structure offers a lot of information about the physical properties of the protein. Being able to accurately predict the chemical shift of a protein is essential in certain areas of molecular dyanamics research such as drug discovery. Currently, there is not an available application that can predict chemical shift of large protein structures in a realistic amount of time, as chemical shift prediction algorithms prove to be too computationally intensive. The VIP students took a chemical shift prediction application called PPM_One, and accelerated it on state-of-the-art GPUs using the directive-based programming standard OpenACC. With the acceleration of the PPM_One application, it can be run on large datasets in a much more reasonable amount of time. When testing several different protein structures on an NVIDIA Titan X GPU, we are seeing a speed-up of roughly 20x.
This work in collaboration with Prof. Juan Perilla from the Dept. of Chemistry and his Ph.D. student Alex Bryer along with CRPL’s PhD student Robert Searles.