Fraunhofer ITWM at Supercomputing 2025

Cutting-Edge Technologies for High-Performance Computers and Artificial Intelligence

Press Release / November 17, 2025

From November 16 to 21, the Fraunhofer Institute for Industrial Mathematics ITWM will be presenting innovative solutions for parallel programming and artificial intelligence at Supercomputing 2025 in St. Louis, USA. Two Fraunhofer spin-offs will be represented at joint booth 3740: ThinkparQ will present its roadmap for the parallel file system BeeGFS. The spin-off UNEEC Systems will demonstrate the STX processor and its system approach for an energy-efficient and cost-effective computing infrastructure. The STX processor accelerates HPC simulations while ensuring easy programming. Researchers from the division »High Performance Computing« will present new developments in the areas of parallel programming, agent-based AI, and accelerated I/O.

GPI-2 – Development of Highly Scalable Parallel Applications

GPI-2 implements the GASPI (Global Address Space Programming Interface) standard, enabling the development of highly scalable parallel applications. Its one-sided, asynchronous zero-copy communication model takes full advantage of the power of today's RDMA-enabled network hardware.

GPI-2 runs on all common communication connections under Linux and offers an API similar to MPI, ensuring compatibility and easy integration. It is particularly effective for scalability-critical parts of HPC codes. At our booth, we will demonstrate how industrial applications achieve excellent scalability with GPI-2.

GaspiLS – A Scalable Iterative Linear Solver for Cfd and Fem Simulations

The iterative solver library GaspiLS has demonstrated exceptional scalability in codes from independent software vendors and industry. Given the widespread use of workstations with over a hundred cores and small HPC installations with thousands of cores, the strong scalability of this solver library is indispensable for technical solutions.

GaspiLS uses highly scalable algebraic multigrid (AMG) and hybrid incomplete LU preconditioners to design the optimal solver for each code. In addition, GaspiLS is open source and can therefore be improved collaboratively. GaspiLS can be seamlessly integrated into any code that uses Message Passing Interface (MPI) or Global Address Space Programming Interface (GPI).

Carme – Open-Source Software Stack for Managing AI Infrastructures

The open-source software stack CARME combines Fraunhofer ITWM's expertise in artificial intelligence and high-performance computing and can be seamlessly integrated into existing HPC systems. Users can focus on their AI development goals without needing experience or knowledge of workload management tools or hardware.

CARME offers web-based access, two-factor authentication (2FA), and integrated container technology, making it a user-friendly and robust tool for managing interactive and batch AI workloads on HPC systems.

Memory Centric Storage (MCS) – Flexible Toolbox for Scalable Storage Management

With Memory Centric Storage, we are developing the European open-source toolbox that combines scalable and domain-specific storage management. Applications access different storage systems via a uniform interface and benefit from zero-copy data transport and task offloading. This allows us to reduce data movement and bring computing processes to where the data is generated – for maximum performance.

The HPC team at Fraunhofer ITWM looks forward to your visit and to stimulating discussions.

Press Release

Fraunhofer ITWM at Supercomputing 2025

Cutting-Edge Technologies for High-Performance Computers and Artificial Intelligence

GPI-2 – Development of Highly Scalable Parallel Applications

GaspiLS – A Scalable Iterative Linear Solver for Cfd and Fem Simulations

More Information

Carme – Open-Source Software Stack for Managing AI Infrastructures

Memory Centric Storage (MCS) – Flexible Toolbox for Scalable Storage Management

Contact Press / Media

Dr. Franz-Josef Pfreundt

Contact Press / Media

M.A. Swenja Broschart