HPC SYSTEMS - PETASCALE PERFORMANCE
GCS offers the most powerful supercomputing infrastructure in Europe, aimed at serving a broad range of scientific and industrial research activities in various disciplines.
Supercomputing at the Leading Edge
Each GCS centre hosts a multi-petaflop supercomputing system, placing all three individual institutions among the most powerful computing centres in the world. With its combined performance, GCS provides the largest supercomputing infrastructure in all of Europe. The system architectures implemented at the three GCS centres are complementary in order to accomodate the broadest range of scientific disciplines.
Hawk of HLRS
Copyright: Ben Derzian, HLRS.
The High-Performance Computing Center Stuttgart (HLRS) hosts an HPE Apollo system named Hawk. The system officially came online in 2020. The machine features 720,896 compute cores and has a theoretical peak performance of 26 petaflops. The system is designed to serve a wide range of sciences, including the life sciences, energy and environmental sciences, high-energy physics, and astrophysics, but places a special empahsis on supporting the computational and scientific engineering communities in academia and industry.
JUWELS of JSC
After the Cluster Module of the Jülich Supercomputing Centre’s (JSC's) HPC system JUWELS (Jülich Wizard for European Leadership Science) went into operation in July 2018, the Booster Module was installed in summer 2020 complementing the cluster system beginning in November 2020. JUWELS consists of 2511 nodes in the Cluster Module and 936 Booster nodes. Cluster nodes are equipped with dual-socket Intel Skylake Platinum 8168 CPUs and InfiniBand EDR interfaces. In addition, 56 Dual Intel Xeon Gold 6148 nodes are equipped with 4 additional NVIDIA Volta V100 GPUs. Each Booster node is equipped with two AMD EPYC Rome 7402 CPUs with 512 GB DDR memory, 4 NVIDIA Ampere A100 GPUs and 4 HDR 200 Gb/s InfiniBand links. JUWELS combines the fat-tree-structured Cluster topology with the Dragonfly+ Booster network in a single high-speed fabric allowing concurrent use of nodes from both modules. The Cluster contributes 12 petaflops to JUWELS massive compute power of 85 petaflops, while the Booster accounts for the majority share with a peak performance of 73 petaflops.
SuperMUC-NG of LRZ
In September 2018, the Leibniz Supercomputing Centre’s (LRZ’s) latest edition to its series of SuperMUC supecomputers was officialy introduced: SuperMUC-NG (“next generation”). With its peak performance of 26.7 Petaflops—an almost fourfold increase of the computing power previously available at LRZ—SuperMUC-NG is currently the fastest supercomputer in Germany. It features an Intel-Lenovo OceanCat platform equipped with 6,336 compute nodes (more than 300,000 compute cores) with Intel Skylake processors and OmniPath interconnects, 700 terabytes of main memory, and 70 petabytes of disk storage.