MS Thesis Defense: Raphael Landaverde

9:00 am on Friday, April 18, 2014
11:00 am on Friday, April 18, 2014
8 Saint Mary's Street, PHO339
GPU Optimizations for a Production Molecular Docking Code -- Date: Friday, April 18, 2014, 9:00am -- 8 Saint Mary's Street, Room 339 -- Advisor: Martin Herbordt (ECE); Committee: Sandor Vajda (BME); Florian Raudies (CAS/CompNet) --- General purpose GPUs have been previously used to accelerated PIPER, an advanced and state of the art molecular docking code. In the current work, we take advantage of modern GPU architectures and new optimizations to accelerate the code even further. Our new version achieves 2x speed up over the fastest parallel CPU version available, with little room for further improvements.