Research Projects

EDIH - European Digital Innovation Hub

Project duration: from October 2022 
Funding: European Union, Ministry of Economic Affairs, Industry and Energy of the State of NRW, German Federal Ministry for Economic Affairs and Climate Protection

The "European Digital Innovation Hub" (EDIH) is a European initiative to promote the innovative capacity of small and medium-sized enterprises. Here, the companies are supported by the partners in the areas of digitalization, machine learning and artificial intelligence as well as high-performance computing (HPC). In the area of HPC, the IT Center, in close cooperation with the Aerodynamic Institute (AIA) and the Chair for the Analysis of Technical Systems (CATS), offers consulting and training in the areas of parallel programming and scientific simulation to enable companies to initially use high-performance computing or to make existing developments in this area more efficient and economical. Further information can be found on the project website.

EE-HPC - Energy Efficient High Performance Computing

Project duration: October 2022 - August 2025
Funding: BMBF Green HPC

The EE-HPC project aims to improve the energy efficiency of HPC systems through automatic system parameter regulation. The project is a joint effort of two  renowned institutions: the German Climate Computing Center (DKRZ) and the High-Performance Computing Center Stuttgart (HLRS). By leveraging the expertise of the other project partners such as Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), Hewlett Packard Enterprise (HPE), and the IT Center of RWTH Aachen, the EE-HPC project is at the forefront of advancing the improvement of HPC data center energy performance.

Together with HLRS, the IT Center is developing a software library for fine-grained energy optimization in parallel MPI and OpenMP domains. Cluster Cockpit, a real-time monitoring and management software, provides valuable insights into power consumption and utilization of the HPC environment. The IT Center is instrumental in interface development and integrates energy-efficient practices into its operations to promote sustainable computing. The ICON application for weather model simulations serves as a test application. Together, the partners are striving for an era of environmentally conscious computing to address scientific research and complex computing challenges in a sustainable manner.
More information is available on the project website

ENSIMA - Energy-Efficient HPC Through Optimized Simulation Methods

Project duration: October 2022 - September 2025
Funding: BMBF Green HPC

The project "Energy-Optimized Simulation Methods for Application-Oriented Computing Problems" - in short "ENSIMA" - is one of the nine third-party funded projects of the BMBF program for "Energy-Efficient HPC (GreenHPC)". The project started on October 1, 2022 and is funded for three years. In addition to RWTH Aachen University as coordinator, the project partners are TU Darmstadt, Forschungszentrum Jülich, Gesellschaft für numerische Simulation mbH, GNS Systems GmbH, SIMCON kunststofftechnische Software GmbH and Technische Hochschule Würzburg-Schweinfurt. All in all, the project involves partners from academia and industry.

The basic goal of the project is to use AI methods to improve the selection of design parameters in production processes and to accelerate the execution time of simulation processes through approximate and heterogeneous computing. Innovative solutions should reduce the number of finite element simulations required, which are used for many engineering problems, such as crash simulations. For the use case of sheet metal forming in the automotive industry, the project partners want to reduce the computation time by 50 percent, which should lead to a 15 percent reduction in the use of steel and thus indirectly to a reduction in both manufacturing-related emissions and the energy required for vehicle production.
More information can be found on the project website.

H2M – Exploiting Heterogeneous Shared Memory Architectures

In the DFG-funded project "Heuristics for Heterogeneous Memory" (H2M), RWTH Aachen University and the French project partner Inria are jointly developing support for new memory technologies such as High Bandwidth Memory (HBM) and Non-Volatile Memory (NVRAM). These technologies are increasingly being used alongside traditional Dynamic Random Access Memory (DRAM) in HPC systems. HBM offers higher bandwidth but smaller size than DRAM. NVM offers greater capacity but is slower than DRAM. Given these differences, the question is how to efficiently use systems with heterogeneous memory and where data should be stored.
 For more information, see the project website.

HPC Software Tools for Debugging and Correctness Analysis

The development of correctness analysis tools and debugging techniques is being driven forward in long-term cooperation with various American research institutions (ANL, LLNL, SNL, LANL, GA Tech and ORNL).

The dynamic MPI correctness analysis tool MUST has been under development since around 2010. As part of the cooperation, the tool was expanded to integrate new MPI functionality, but also to better support user-driven use cases. The Archer tool, which recognizes data races in OpenMP applications, has been under development since 2014. As part of the cooperation, the tool was expanded to cover new OpenMP functionality and to integrate new analysis techniques. The tool has been integrated into the LLVM project since 2020. The cooperation makes it possible to keep the two tools ready for production so that they are available to the HPC community for the (further) development and parallelization of HPC applications.

To better support the debugging of highly parallel HPC applications, we are developing new debugging techniques as part of the cooperation with the aim of improving debugging support for OpenMP and MPI applications. Significant contributions to the implementation of the OpenMP Debugging Interface (OMPD) in LLVM have emerged from this co-operation.

Another aspect is the detection of errors in OpenMP applications when dealing with hardware accelerators such as graphics cards. As part of the co-operation, various techniques for detecting such errors are being investigated. This has resulted in prototype tools such as Arbalest and TSan-SPD3.

HPC.NRW

Project duration: April 2019 - December 2024
Funding: Ministry of Culture and Science of the State of North Rhine-Westphalia (only german)

The HPC.NRW competence network offers an extensive range of diverse HPC infrastructures plus a number of thematic clusters for low-threshold training, consulting and coaching services. The aim is to make effective and efficient use of high-performance computing and storage facilities and to support scientific researchers of all levels, particularly researchers at the post-graduate and post-doctoral levels. Regular, intensive exchanges take place between the sites of the competence network on the service and support for HPC users in NRW, software and operation, as well as HPC usage. More information can be found on the project page.

IT-ZAUBER - Digital Twins for Energy Efficient Data Centers

Project duration: September 2022 - August 2025
Funding: BMBF Green HPC

In the current BMBF project IT-ZAUBER, RWTH Aachen University, TU Dresden and ROM-Technik GmbH & Co. KG, together with the associated partners Sachsen Energie AG and ICT Facilities GmbH, are developing a digital twin as a digital image of the data center that captures its properties, condition and behavior through models, information and data. The application possibilities as a digital tool for the planning and operation of data centers are to be demonstrated and evaluated exemplarily at the two HPC centers of the RWTH and the TUD. The digital twin will be aware of the condition and behavior of the IT infrastructure as well as the cooling system and its integration into the other supply structures and will be used for condition evaluations and target value determinations, thus going far beyond the state of current operating concepts. The project is one of nine collaborative research projects in the field of energy-efficient HPC (GreenHPC) with three years of funding from the BMBF (FKZ 16ME0614).

NHR4CES

In NHR4CES (National High Performance Computing Center for Computational Engineering Science), RWTH Aachen University and Technical University Darmstadt join forces to combine their strengths in HPC applications, algorithms and methods, and the efficient use of HPC hardware. The goal is to create an ecosystem combining best practices of HPC and research data management to address questions that are of central importance for technical developments in economy and society. More information can be found on the project website. 

Performance, Optimization and Productivity - POP3

Project duration: January 2024 - December 2026
 

The [Centre of Excellence on Performance Optimization and Productivity (POP CoE) supports HPC application developers in science and industry in the performance-related analysis and optimization of their codes. Such analyses focus in particular on the growing heterogeneity and complexity of hardware (accelerators, increasing number of cores, deeper memory hierarchies). In the third phase (POP3) of the project, the extremely successful work of the previous two phases will be continued. The project focuses on three pillars consisting of services, users and co-design. In the POP services, an objective review of code performance and scaling is carried out and, based on the results, causes for inefficient code parts are identified and suggestions for improvement are developed. POP3 also offers higher-value services, such as proof-of-concept studies and energy efficiency and correctness analyses. The service concentrates on highly scalable applications, i.e. flagship applications from other centers of excellence (CoEs), but is also open to other HPC users. The user pillar focuses on user training (offering workshops and tutorials) as well as customer development and customer satisfaction. The co-design pillar aims to improve and further develop the POP tools and methodology for the objective evaluation of code performance on other hardware platforms (accelerators, modern processor architectures).

Further information can be found on the project website.

Research on AI- and Simulation-Based Engineering at Exascale (RAISE)

Project duration: June 2022 - May 2024
Funding: EU’s Horizon 2020

The “Research on AI- and Simulation-Based Engineering at Exascale” (RAISE) project is a European Center of Excellence in Exascale Computing funded by the European Commission under the Horizon 2020 Framework Programme. A consortium of 13 partners from academia and industry works together on the convergence of traditional HPC and innovative AI techniques along various use cases from different domains. Use-cases involve compute-driven applications, such as AI for turbulent boundary layers, as well as data-driven applications like event reconstruction and classification at the CERN High-Luminosity Large Hadron Collider. Results obtained with these use-cases are integrated into a Unique AI Framework that contains the trained models and documentation how to use the developed AI techniques on current Petaflop and future Exaflop HPC systems. 
More information can be found on the project website.

targetDART - Dynamic, Adaptive and Reactive Distribution of Compute Tasks on Heterogenous Exascale Architectures

Project duration: October 2022 - September 2025
Funding: BMBF SCALEXA

In the BMBF project targetDART, the Technical University of Munich (TUM), the High Performance Computing Center Stuttgart (HLRS) and RWTH Aachen University, represented by the IT Center, are developing a task-based approach for highly scalable simulation software that compensates for unpredictable load imbalances on heterogeneous exascale systems by means of dynamically adaptive and reactive distribution of computational tasks between computational nodes.

The BMBF project Chameleon, which has already been completed, serves as the basis for targetDART and provides valuable insights and results on dynamic task migration of tasks between nodes based on a library/API.

The project employs the OpenMP target construct for GPU utilization and MPI, particularly the new 4.0 standard, for efficient communication between nodes. The migration approach, particularly for GPUs, will be further explored and clarified. The project will also apply its approach to SeisSol, a dynamic earthquake and seismic wave simulator, and ExaHyPE, a solver for hyperbolic partial differential equations, for comprehensive testing and evaluation.

The IT Center focuses on OpenMP, target constructs, and CPU-GPU migration, while HLRS prioritizes MPI and node migration optimization. TUM's emphasis lies in optimizing the two applications. Together, the project's dynamic approach and collaboration will drive innovation in high-performance computing.
For more information, visit the project website.

Virtual Institute - High Productivity Supercomputing

Sponsored by the Helmholtz Association of German Research Centers the Virtual Institute - High Productivity Supercomputing (VI-HPS) aims at improving the quality and accelerate the development process of complex simulation programs in science and engineering that are being designed for the most advanced parallel computer systems. The IT Center of the RWTH Aachen University is focussed on improving the usability of the state-of-the-art programming tools for high-performance computing developped by the partner institutions. More information can be found on the project website.