high performance computing on graphics processing units: hgpu.org

Epidemiology computation models are crucial for the assessment and control of public health crises. Agent-based simulations of pandemic influenza are useful for forecasting the infectious disease spreading in order to help public health policy makers during emergencies. In such emergencies decisions are required for public health preparedness in cycles of less than a day, and the agent-based model should be adaptable and tractable for quick and simple calibration with low computational overhead. GPU accelerated computing involves the use of a graphics processing unit (GPU) in combination with the CPU to perform heterogeneous computing by offloading a compute-expensive portion of the program to the GPU while the remaining program is running on the CPU. This thesis modifies former models considerably, explores the performance of a low-complexity agent-based model for pandemic simulations when accelerated by multiple GPUs on a single node computer. In this thesis, we demonstrate the utilization of the hardware environment and software tools and discuss strategies for adapting agent-based simulation to multiple GPUs. We further compare the performance of simulations using two GPUs or four GPUs with the sequential execution on the CPU, in terms of time and speedup. The multi-GPU implementations exhibit great performance and support populations with up to 100 million individuals.