Filesystem integration for a GPU based Java virtual machine

Modern computers utilize many accelerator devices alongside traditional CPUs. These devices provide additional performance or functionality to the system for specialized workloads. We specifically investigate graphics processing units (GPUs), which have become very prolific in enterprise and consumer computers. GPUs can enable demanding graphics-based workloads, but they can also be used to perform more flexible general purpose computation (GPGPU). Currently, the most popular GPGPU workloads are related to machine learning such as training a neural network. While this may be the case, there are many other highly parallel workloads that can take advantage of the GPU architecture. This project focuses on extending GVM, a Java interpreter that enables running native Java code on GPUs, to provide direct filesystem access.