Target group
- Scientific research, companies, public institutions
Your requirements
- Particularly high data protection and confidentiality requirements
- Use of high-performance computing (HPC) resources
- Need for GPU-based systems for training and inference tasks
- High availability inference platform
- Guaranteed time-to-completion
Our offer
For users with particularly sensitive data, we also offer the applications and services of the AI service center on a secure, isolated partition that is specially hardened. This includes, for example, a GPU-based HPC system with current NVIDIA A100 and H100 GPUs provided for training and inference tasks in research, development and engineering, respectively. The compute nodes are interconnected via InfiniBand or RoCE networks and with the storage resources that are also provided. Appropriate network segmentation is used to prevent external access to the data. Software, models and data can be installed in self-service or integrated via the KISSKI catalogs. The system can be used either in direct access or as a technical basis for the KISSKI services.Requirements
For direct use of the computing resources, a current SSH client is required, as well as the client software for the secure HPC environment. Individual prerequisites apply for indirect use of the computing resources through the services offered at KISSKI.
