The goal of this DURIP proposal is to build a GPU based compact supercomputer testbed called ASPIRE that would support the ongoing ONR MURI research on advanced persistent threats, with project name ADAPT, that is led by the University of Washington Team. ADAPT is a data-driven MURI project that aims to model the interactions between the adversary and the systems. The detailed system model and the interaction logs are created by MURI team members from Georgia Tech. The data volume is quite large requiring computational capabilities that would be supporting the efforts. In addition, the UW team has been investigating adversarial machine learning in real application data sets such as images and videos. Each of these data sets typically require several days of computations on regular CPU machines. The GPU based systems are built with such data intensive machine learning applications in mind and are well suited to handle needed computations.