HLRS Launches Future Computing Group to Test 3 New HPC Architectures as GPU Costs Rise
Updated
Updated · Scientific Computing World · May 25
HLRS Launches Future Computing Group to Test 3 New HPC Architectures as GPU Costs Rise
2 articles · Updated · Scientific Computing World · May 25
HLRS has set up a Future Computing Group to evaluate quantum, neuromorphic and alternative accelerator systems for future high-performance computing workloads beyond CPUs and GPUs.
The move reflects mounting physical and economic limits on conventional scaling, with GPU-based supercomputers becoming more expensive and energy-intensive as HPC systems grow more heterogeneous.
Johannes Gebert’s team will work with hardware developers to measure how new architectures handle algorithms, workflows and software environments used in scientific and industrial computing.
Openchip is among the first partners, with HLRS testing the Barcelona company’s RISC-V vector accelerators for large-scale HPC jobs, sustainability gains and easier application porting than GPUs.
The program fits HLRS’s longer-term roadmap after bringing the Hunter supercomputer online in 2025 and ahead of its planned exascale Herder system in 2027.
As NVIDIA's dominance grows, can Europe's pivot to open-source RISC-V create a truly competitive supercomputing ecosystem?
With future supercomputers mixing quantum, AI, and RISC-V, how will scientists actually program these radically new hybrid machines?
HLRS Future Computing Group: Pioneering Hybrid and Sustainable HPC Architectures for Europe’s Next-Generation Supercomputers
Overview
HLRS has launched its Future Computing Group to proactively address the evolving landscape of high-performance computing. This group explores and evaluates technologies beyond traditional CPUs and GPUs, aiming to identify the next generation of infrastructure that will support advanced algorithms, software, and heterogeneous computing workflows. By bridging the gap between innovative technology companies and the scientific community, HLRS ensures that future HPC systems are tailored to answer complex research questions. Recognizing the growing demand from scientists for faster computing, HLRS is committed to propelling scientific computing to new levels of capability.