Data Analysis and Visualization
CSCS provides core expertise to help its users with data exploration and visualization, based on four main activities with the aim of providing both the hardware and the software resources to achieve effective data understanding:
- to support efficient simulations with large I/O loads, we interact with the scientists to adopt the most appropriate I/O library and to optimize its usage, both in writing, but also in reading, since parallel I/O are often a bottleneck of high-scale visualization. Selective I/O, such as out-of-core processing, level-of-details, ghost data support, streaming and parallel support can be examples we provide to our community. We can also assist the users in migrating to in-situ visualization, to overcome the penalty of large I/O to disks
- we provide the expert tuning of parallel visualization tools such as ParaView or VisIt to efficiently load-balance their computations, and make them operate in a mix of batch and interactive modes across remotely located computers
- we share our many years of experience in multiple application fields and will guide the users in selecting the most appropriate visualization and image generation techniques for their respective fields
- we guide the users in adopting newer paradigms of data exploration, data analytics, and classical data analysis, to find information hidden in their data to enhance understanding and discovery.