1: \begin{abstract}

2: %Containers have been long used in the cloud landscape as they provide benefits such as security, fault tolerance, elasticity and ease of deployment. In the hardware landscape, cloud-optimized processors with ARM-based microarchitecture brings performance at a lower energy point. Despite the benefits, these technologies are not widely adopted in HPC environments although there is a growing trend in their usage. In this paper, we analyse how six different HPC workloads perform in three different ARM-based microarchitectures, and characterize the existing overhead from the containers by comparing with the baremetal performance.

3: 

4: Recent development in lightweight OS-level virtualization, containers, provides a potential solution for running HPC applications on the cloud platform. In this work, we focus on the impact of different layers in a containerized environment when migrating HPC containers from a dedicated HPC system to a cloud platform. On three ARM-based platforms, including the latest Nvidia Grace CPU, we use six representative HPC applications to characterize the impact of container virtualization, host OS and kernel, and rootless and privileged container execution. Our results indicate less than 4\% container overhead in DGEMM, miniMD, and XSBench, but 8\%-10\% overhead in FFT, HPCG, and Hypre. We also show that changing between the container execution modes results in negligible performance differences in the six applications.

5: 

6: \keywords{Cloud and HPC Convergence \and Containers \and ARM \and Performance}

7: \end{abstract}

8: