Masterarbeiten zu Computing
| Thema: | GPU support for the Grid |
| Zusammenfassung: | GPUs play an increasingly important role in High Energy Physics (HEP). Therefore, KIT provides GPUs accessible to various HEP collaborations via the distributed computing infrastructure, the so-called Grid. Currently, there are only initial concepts. You would contribute to implementing the GPU support in the Grid on a collaboration-wide scale. It can be complementary to the PhD or MSc thesis. |
| Sie lernen kennen: | Batch System HTCondor, GPU usage, Grid infrastructure |
| Referent: | Prof. Dr. Torben Ferber |
| Ansprechpartner: | Dr. Matthias Schnepf |
| Letzte Änderung: | 15.10.2024 |
| Thema: | Reinforcement Learning for Dynamic Resource Management |
| Zusammenfassung: | KIT develops concepts and software to integrate opportunistic computing resources transparently. The decision of how many resources of which type has to be provided for efficient usage is complex. Reinforcement learning, where an artificial neural network learns itself, can increase resource efficiency and reduce resource specific parameter tuning. The implementation and study of reinforcement learning can complement the PhD or MSc thesis. |
| Sie lernen kennen: | Batch System HTCondor, bash, python, machine learning, Grid infrastructure |
| Referent: | Prof. Dr. Torben Ferber |
| Ansprechpartner: | Dr. Matthias Schnepf |
| Letzte Änderung: | 15.10.2024 |