CESGA’s computing resources help scientists to diagnose the dry eye syndrome
Blog written with the support of Jorge González-Domínguez, who is a researcher at the University of Coruña
Dry eye syndrome is an eye disease that can hinder many daily activities, from driving to working with computers. The diagnosis usually involves building tear film maps that can be automatically generated via existing applications.
However, medical experts often complain about the time needed for its completion (several minutes).
Jorge González-Domínguez and his team of researchers at the University of Coruña (Spain), in collaboration with INESC TEC – INESC Technology and Science (Portugal), found a method to accelerate the production of tear film maps on distributed-memory clusters. This new approach modifies the original algorithm for drawing tear film maps by using Message Passing Interface (MPI) for inter-node communications and the multithreading support available in the C++11 standard for intra-node parallelization.
Message Passing Interface (MPI) is a standardised and portable message-passing system designed to function on a variety of parallel computing architectures.
The method was tested on 32 nodes of CESGA’s Finis Terrae II supercomputer. The results, published in Computer Methods and Programs in Biomedicine, show that maximum runtime is reduced from almost two minutes using the previous only-multithreaded approach to less than ten seconds using this method. Prior to this work, the original sequential C++ implementation (without multithreading) needed up to half an hour to generate some heavy tear film maps.
Runtime reduction can extend the use of tear film maps among medical experts and help to diagnose the dry eye syndrome.