Helmholtz Zentrum München (HGMU) is the German Research Center for Environmental Health. The HGMU investigates important common diseases that develop from the interaction of lifestyle, environmental factors and personal genetic background, with particular emphasis on diabetes mellitus and chronic lung diseases.
The challenge was to develop new algorithms and full system administration of the institute’s compute cluster (currently with 1,700 cores) utilized by its 35 employees. The facility was using an open source version of Grid Engine, version 6.2u5.
The open source version was not in alignment with technological advancements such as working with many core machines or having a scheduler that recognizes traditional GPUs.
While the investment of new software licenses was a financial commitment they have identified specific advantages in migrating to Univa Grid Engine. The deployment was painless and the software was configuration stable.
They now have an environment that schedules workloads to GPUs and they get to take advantage of scheduling 45% more jobs – especially the more important parallel jobs – in any given day.
Download the complete case study at: http://bit.ly/1GbtqSl