HPC machine learning computer vision containers in HPC clusters
Issue Date:
5-Nov-2020
Publisher:
Institute of Mathematics and Informatics Bulgarian Academy of Sciences, Association for the Development of the Information Society
Citation:
Proceedings of the National Conference on "Education and Research in the Information Society", Plovdiv, November, 2020, 057p-063p
Series/Report no.:
ADIS;2020
Abstract:
Docker containers are not the first choice in the field of high performance computing
(HPC) due to the need for escalation of the privileges in the implementation of the containers
as well as the difficult one integration with batch systems. Singularity is a better alternative to
Docker containers in HPC developed by the Berkeley Laboratory (Lawrence Berkeley National
Laboratory). The performance of containerized applications and the ability to set up and
configure parallel applications with multiple dependencies on external (third party) software
packages is assessed for the following cases: one compute node is only used (32 cores);
multiple software packages in one container for classification of chest X - ray images of patients
with SARS-CoV-2 https://github.com/lindawangg/COVID-Net using the platform for Tensorflow
open source machine learning and computer vision library OpenCV.
Description:
Report published in the Proceedings of the National Conference on "Education and Research in the Information Society", Plovdiv, November, 2020
