SC26: Fraunhofer ITWM, High Performance Computing is Becoming Smarter and More Energy-Efficient

At SC26 in Chicago, the Fraunhofer Institute for Industrial Mathematics ITWM presents the latest developments in HPC. Together with co-exhibitors ThinkParQ GmbH, featuring BeeGFS, and the Swiss deep-tech start-up Qambria AG, the institut, showcases solutions for heterogeneous HPC systems, agent-based AI, innovative storage technologies, energy-adaptive data centers, and Quantum Computing, at Booth 2516.From November 15 to 20, 2026, the international HPC-Community gathers at SC26 in Chicago. The exhibition opens from November 17th to 19th at McCormick Place. Fraunhofer ITWM provides its expertise in mathematical modeling, simulation, and optimization to develop solutions for demanding applications in industry and science.

The focus of the exhibition is on three key challenges in HPC: the efficient use of energy and data, the intelligent automation of complex computational processes, and the integration of new computer architectures. To address these challenges, Fraunhofer ITWM considers not just individual components, but the entire HPC-System.

“In the future, HPC will not be determined solely by fast processors. Intelligent software, efficient data utilization, mindful energy management, and the ability to seamlessly integrate new technologies such as quantum computing are crucial. This is precisely the combination we are working on at Fraunhofer ITWM,” said Dr. Jens Krüger, head, analytics and computing division, Fraunhofer ITWM.

An Overview of the entire HPC-System
Modern HPC-Systems combine different Processors, Memory Technologies, and Computer Architectures. For applications to make efficient use of such heterogeneous Infrastructures, numerous components must be coordinated with one another. Fraunhofer ITWM develops Solutions across the entire Computing-Stack: from Algorithms and Application modeling to Processors, Memory, and high-speed Data Connections, all the way to Data Center operations.

Not only does the most powerful computing performance matter, furthermore equally important are fast Data Access, Scalable Applications, and resilient Operation. The institute develops customized Solutions and supports Companies and Research Institutions with vendor-neutral benchmarking and consulting services for the deployment and operation of HPC-Systems.

Agent-based AI automated Simulation-Workflows
A key feature of the exhibition is agent-based, whom, can independently support, coordinate, and link individual steps in complex simulation and analysis processes. The Fraunhofer ITWMs approaches range from the configuration and execution of a simulation to the analysis and presentation of the results. This shortens the path from computer-aided experimentation to actionable insights.

Locally deployable solutions play a significant role in this process. Companies and research Institutions can deploy agent-based AI directly within their own IT and HPC-Infrastructure without having to transfer sensitive data to external cloud services. This creates transparent AI workflows that ensure data sovereignty for industrial or scientific applications.

Efficient Memory Management for Data-Intensive Applications
As computing power increases, so do the demands placed on Storage-Technologies. Storage-Resources are becoming more expensive and are increasingly rising the cost and performance of modern HPC-Systems. In Data-Intensive Applications, performance therefore depends not only on the processors, but also on how quickly and efficiently the systems store, retrieve, and process data.

With “Memory Centric Storage” (MCS), Fraunhofer ITWM is developing a European open-source toolbox for scalable and Application-Specific Storage Management. Applications access different storage technologies via a unified interface. zero-copy data transfer avoids unnecessary copy operations, while task offloading shifts computational tasks closer to where the data is generated. This reduces data movement, makes resource usage more transparent, and improves the performance of data-intensive applications.

“FlexLab”: Data Centers are becoming part of the Energy System
Furthermore, the energy needs of data centers are receiving greater attention. In the project “FlexLab” Fraunhofer ITWM is developing new concepts for energy-efficient and flexible data centers. The researchers are investigating how computing power can be intelligently adapted to the available energy supply.

In the future, computing operations and technical infrastructure are expected to be able to respond more dynamically to conditions in the energy system. As a result, data centers would no longer merely consume energy but could also contribute to the integration of renewable energy and the stabilization of power grids. The project thus combines energy efficiency and energy flexibility with the goal of strengthening Europe’s digital sovereignty.

A Bridge Between HPC and Quantum Computing
Another topic at the fair is the integration of quantum processors into classical HPC-Infrastructures. Together with the Swiss deep-tech start-up Qambria AG, Fraunhofer ITWM is demonstrating how these two computing worlds can be connected. Qambria is developing a vendor- and modality-neutral software and control layer for this purpose, which seamlessly integrates Quantum Pocessors into classical HPC Environments.

Qambria is exhibiting for the first time alongside Fraunhofer ITWM at SC26. Among other highlights, the focus is on solutions for the real-time control of quantum processors and error correction. These are key prerequisites for the reliable deployment of future Quantum Computers.

Entitled “Quantum Hybrid3” the Fraunhofer ITWM is also showcasing approaches to hybrid quantum AI. Quantum Hybrid AI combines the computing power of Quantum Processors with the reliability and scalability of classical Supercomputers and Graphics Processing Units. Depending on the task, calculations can be performed on the most suitable hardware. »Quantum computers will not replace classical HPC-Systems, but rather complement them in specific ways. That is why we are developing technologies that bridge the two worlds and enable the practical use of hybrid applications«, says Dr. Jens Krüger.

Automatically Discovering and Optimizing Quantum Architectures
Which quantum circuit is best suited for a specific task? Fraunhofer ITWM answers this question with Quantum Architecture Search (QAS). The technology automatically searches for and optimizes suitable quantum architectures, taking into account both the accuracy of the results and the required resource expenditure. Switch structures and input encodings are learned directly from sample data. This allows new Quantum Applications to be developed more efficiently and adapted to different types of Quantum Hardware.

“QUALLE” ensures transparency in Quantum-AI-Benchmarking
The Fraunhofer ITWM is also introducing “QUALLE” a new product for benchmarking quantum AI methods. The goal is to systematically, transparently, and comprehensibly evaluate the performance of quantum ML applications. This enables the assessment of which tasks quantum methods already offer added value for and how they compare to classical solutions.
The research conducted at the institute ranges from quantum ML and the Optimization of Quantum Networks to agent-based Circuit gen and agent-based Benchmarking.

New GPI-2 Release Improves Communication in Large HPC Systems
With the new GPI-2 Release 1.8.0, Fraunhofer ITWM is presenting a further development of its programming model for efficient and scalable HPC applications at SC26. GPI-2 enables communication and computations to be executed largely in parallel, thereby allowing even large distributed simulations to be processed efficiently. The new version supports additional network technologies and improves the launch of large computing jobs with many processes.

Another new feature is a runtime analysis function: applications can be analyzed during execution without having to recompile the program code first. The results can be evaluated directly using common visualization tools. This makes it easier to identify performance bottlenecks in large HPC applications and to optimize the programs in a targeted manner.

Gaspils Speeds up Time-Dependent Simulations
The “GaspiLS” solver library, developed at Fraunhofer ITWM, has also been further refined. In time-dependent simulations, similar systems of equations must be solved over many consecutive computational steps, for example in flow or reservoir simulations. GaspiLS specifically exploits this similarity and can reuse previously calculated structures in subsequent computational steps. This reduces the computational effort and, in particular, shortens the time-consuming preparation of the individual solution steps.

A benchmark from a reservoir simulation involving approximately one million cells demonstrates the magnitude of this effect: Across the tested process numbers, GaspiLS achieved speeds up to twice as fast as the comparison methods used. For complex transient simulations, this opens up the possibility of using computational resources more efficiently and delivering results more quickly.

In addition to these key areas, Fraunhofer ITWM provides further insights at SC26 into current developments for high performance and scalable HPC systems, including the “STX processor”. The technologies presented demonstrate how algorithms, software, communication, memory, and hardware can be specifically coordinated to make efficient use of heterogeneous HPC systems.

Together with co-exhibitors ThinkParQ GmbH and Qambria AG, Fraunhofer ITWM provides a comprehensive overview of current and future computing architectures at Booth 2026. The exhibition presentation covers a wide range of topics, from high performance memory and simulation solutions to locally deployable AI agents and energy-adaptive data centers, all the way to the integration of quantum processors into classical HPCs.