Treffer: Loop then task: Hybridizing OpenMP parallelism to improve load balancing and memory efficiency in continental-scale longest flow path computation.

This study presents a new OpenMP parallel algorithm for Memory-Efficient Longest Flow Path (MELFP) computation for large-scale hydrologic analysis. MELFP hybridizes loop-based and task-based parallelism to improve load balancing and eliminates intermediate read-write matrices to optimize memory usage. Its performance remained insensitive to the threshold parameter for switching from looping to tasking. Compared to the benchmark algorithm, MELFP achieved a 66 % reduction in computation time while increasing CPU utilization by 33 %. Its 79 % lower peak memory usage enables processing larger datasets. These results suggest that MELFP is a fast and memory-efficient solution for longest flow path computations across a large number of watersheds, particularly in high-performance computing environments where rapid execution is prioritized over lower CPU utilization. MELFP's additional ability to compute longest flow paths for individual subwatersheds provides added benefits for detailed and localized hydrologic modeling. [Display omitted] • The Memory-Efficient Longest Flow Path (MELFP) algorithm is introduced. • It hybridizes loop-based and task-based OpenMP parallelism to improve load balancing. • It does not require intermediate read-write matrices to reduce memory usage. • It achieved a 66% reduction in computation time using 79% less peak memory. • It increased CPU utilization by 33%, making it suitable for HPC environments. [ABSTRACT FROM AUTHOR]

Copyright of Environmental Modelling & Software is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)