The intent of NAIC is to provide a High-Performance Computing (HPC) infrastructure for the exploration of AI methods in various domains for researchers, PhD students, educators, entrepreneurs, startups, and SMEs. To make the most out of the NAIC infrastructure, and make it usable for a wide range of users, the infrastructure should be used efficiently. Resources should only be allocated when and where they are needed. To achieve this, however, NAIC has to gain an understanding of typical computational needs and the variety of users’ experience levels.
NAIC's initial proposal comprised the additional acquisition of computing hardware - mainly GPUs - to improve the offered computing capabilities. Since SIGMA2 was already able to secure a large additional number of GPUs, NAIC now takes the opportunity to use its budget for improving the infrastructure at various other levels. Firstly, handling data in a user's workload will become easier and faster by improving storage solutions. Secondly, NAIC will establish mechanisms to classify workload types and usage patterns, so that NAIC can offer better fitting resources with respect to user workload requirements and suggest how to improve computing performance.
As a result, more users should be able to use the infrastructure at the same time, and jobs will run faster.