High-performance computing for super-resolution microscopy on a cluster of computers

TL;DR

This paper develops a highly efficient, scalable parallel implementation of the MUSICAL super-resolution microscopy algorithm on a 256-core computer cluster, significantly reducing processing time for practical applications.

Contribution

A novel parallel MUSICAL algorithm optimized for clusters, achieving high speed-up and efficiency using advanced parallel programming techniques and libraries.

Findings

Achieved 240.29x speed-up on 256-core cluster

Completed super-resolution microscopy within 10 seconds

Efficiency of 93.86% on the cluster

Abstract

Multiple signal classification algorithm (MUSICAL) provides a super-resolution microscopy method. In the previous research, MUSICAL has enabled data-parallelism well on a desktop computer or a Linux-based server. However, the running time needs to be shorter. This paper will develop a new parallel MUSICAL with high efficiency and scalability on a cluster of computers. We achieve the purpose by using the optimal speed of the cluster cores, the latest parallel programming techniques, and the high-performance computing libraries, such as the Intel Threading Building Blocks (TBB), the Intel Math Kernel Library (MKL), and the unified parallel C++ (UPC++) for the cluster of computers. Our experimental results show that the new parallel MUSICAL achieves a speed-up of 240.29x within 10 seconds on the 256-core cluster with an efficiency of 93.86%. Our MUSICAL offers a high possibility for real-life applications to make super-resolution microscopy within seconds.