SCALABLE - D3.3 Implementation and final report about scheduling and load-balancing
Raphaël Kuate · Zenodo (CERN European Organization for Nuclear Research) · 2023
The main objective of SCALABLE for CSGROUP is the improvement of LaBS deployment in bigger clusters of thousands of cores, achieved by a transfer of performance technology from WaLBerla. Therefore, in the earlier work packages 3.1 [1], 3.2 [2] and 3.5 [3] many improvements were made to LABS scalability. In the previous reports on work package 3.5 and 3.2 we introduced works on LaBS targeting GPU clusters. In this report, we present the development of a prototyping GPU version of LABS