Implementing Scientific Simulations on GPU-accelerated Edge Devices

In this paper, we investigate the potentials of the state-of-the-art edge devices in the context of scientific computing. We implement one of the major scientific applications - stochastic simulation in biochemistry - on a small cluster of modern, GPU-accelerated edge systems. By comparing the performance of the edge implementation with that of a real-world stochastic simulation software package on a multi-core desktop system, we evaluate the computational capability and estimate the usefulness of the modern hardware-accelerated edge devices.