Home | Sitemap | deutsch  | Legals | KIT

High Performance Computing

Multi- and many-core supercomputing

High-performance computing (HPC) is a filed of computation, where big memory and processing power is required. Here especially parallel processing of computations plays an important role. Programming interfaces like MPI allowing development of applications for HPC cluster architectures geared to massively parallel processing. Such computing clusters are optimized for use of possibly fast interconnections in extremely short communication time between computing units. Here are high-speed transfer technologies like InfiniBand are used. Applications using such a architectures are a need to be adapted to special software environments and usually need a communication network with low latency and high throughput. Additionally the memory usage could be very high. To support scientific communities in using high-end supercomputing resources, SCC has established a new type of domain-specific research and support structure: the Simulation Laboratories.

 

Projects

 

bwHPC-C5: Coordinated Compute Cluster Competence Centers

In response to the increasing significance of computational science and engineering in science and research, the State of Baden-Wuerttemberg has developed a consistent concept “bwHPC” for high performance computing covering HPC support at all levels. As a statewide accompanying project “bwHPC-C5” (Coordinated Compute Clusters Competence Centers) has been coordinating since July 2013 the federated support of the users and the related measures and activities in the State of Baden-Wuerttemberg. Aim of the project is the organisation of federated HPC competence centres as an interface between science and high performance computing.

 

High Throughput Protein Structure Prediction on
Hybrid and Distributed High Performance Architectures (HPC-5)

The great challenge in the coming years is the development of high throughput methods for the quantitative prediction of the three-dimensional structure of proteins with a low sequence similarity to structure-resolved proteins. The project HPC-5 being funded by the “Baden-Württemberg Stiftung” (foundation of the State of Baden-Wuerttemberg) will contribute substantially to the solution of this problem. In cooperation with the Institute of Nanotechnology novel models and algorithms are combined with established techniques from bioinformatics in order to model the structure of large proteins of biological and pharmaceutical relevance. The deployment of efficient algorithms on HPC architectures and the utilisation of new process architectures (such as GPU) by adopting capability computing can help to master unsolved problems in the field of life sciences.

Multi-scale materials modelling on
high performance computer architectures (MMM@HPC)

In this FP7 project expert groups from the field of materials research (code developers), HPC resource providers and users from industry bundle their competences in order to develop and provide an integrated e-infrastructure for multi-scale materials modelling ((MMM@HPC). The open MMM@HPC infrastructure will allow the integration of existing software modules into customisable high performance protocols and workflows as well as the application on distributed computer resources. MMM@HPC unites academic and industrial scientists in the field of computational science and cooperates with other European projects and organisations like PRACE. The project is coordinated by SCC and the Institute of Nanotechnology (INT).