Zynq SoC based Lattice-Boltzmann Simulation Environment

Cerebral aneurysm is a weakness in a blood vessel that may enlarge and bleed into surrounding area, which is a lifethreatening condition. Therefore, early and accurate diagnosis of aneurysm is highly required to help doctors to decide the right treatment. The HemeLB is a massively parallel lattice-Boltzmann simulation framework that allows surgeons to visualise the simulation results. However, the HemeLB simulation framework is originally designed for high performance computer (HPC), which is not user friendly for the end users. In this paper, we present a solution for designing and implementing HemeLB on cost efficient embedded platforms, in order to allow the HemeLB simulation framework to be potentially implemented in the local environment of hospital. The proposed work has been developed using a Zynq system-on-chip (SoC), which is a heterogeneous multi-processor system-on-chip (MPSoC) platform widely used in high performance embedded applications. In addition, a comprehensive evaluation for the implementation using the data extracted from real patient with cerebral aneurysm segmentation is also reported in this paper. The results demonstrate that the proposed implementation is able to support the HemeLB framework on a low-cost embedded platform and achieving a maximum performance of 215,751 sites updates per second with 2 cores.